Authors: Mikael von Strauss
Abstract: Under real-analytic assumptions on decoder-only Transformers, recent work shows that the map from discrete prompts to last-token hidden states is generically injective on finite prompt sets. We refine this picture: for each layer $\ell$ we define a collision discriminant $\Delta^\ell \subset \Theta$ and injective stratum $U^\ell = \Theta \setminus \Delta^\ell$, and prove a dichotomy -- either the model is nowhere injective on the set, or $U^\ell$ is open and dense and every $F^\ell_\theta$ is injective. Under mild non-singularity assumptions on the optimizer and an absolutely continuous initialization, generic injectivity persists along smooth training trajectories over any fixed horizon. We also treat symmetry groups $G$, showing that discriminants and injective strata descend to the quotient $\Theta/G$, so injectivity is naturally a property of functional equivalence classes. We complement these results with an empirical study of layerwise geometric diagnostics. We define a separation margin and a co-Lipschitz (lower Lipschitz) constant between prompt space and last-token representation space, estimated via nearest-neighbor statistics on large prompt sets. Applying these diagnostics to pretrained LLaMA-3 and Qwen models, we study behavior across layers, sequence lengths, model scales, and 8- and 4-bit activation quantization. On our sampled prompts we see no collisions in full precision or at 8 bits, while 4-bit quantization induces a small number of collisions and markedly shrinks co-Lipschitz estimates. For a small GPT-2 trained from scratch, normalized metrics remain stable over training. Overall, the results suggest that Transformer representations are generically and persistently injective in the continuous-parameter idealization, while their practical invertibility can be probed using simple geometric diagnostics.
Authors: Yifan Li, Qin Li, Min Zhang, Min Zhang, Peixin Wang
Abstract: Assessing the reasoning ability of Large Language Models (LLMs) over data remains an open and pressing research question. Compared with LLMs, human reasoning can derive corresponding modifications to the output based on certain kinds of changes to the input. This reasoning pattern, which relies on abstract rules that govern relationships between changes of data, has not been comprehensively described or evaluated in LLMs. In this paper, we formally define this reasoning pattern as the Derivation Relation (DR) and introduce the concept of Derivation Capability (DC), i.e. applying DR by making the corresponding modification to the output whenever the input takes certain changes. To assess DC, a systematically constructed evaluation framework named DEVAL is proposed and used to evaluate five popular LLMs and one Large Reasoning Model in seven mainstream tasks. The evaluation results show that mainstream LLMs, such as GPT-4o and Claude3.5, exhibit moderate DR recognition capabilities but reveal significant drop-offs on applying DR effectively in problem-solving scenarios. To improve this, we propose a novel prompt engineering approach called Derivation Prompting (DP). It achieves an average improvement of 15.2% in DC for all tested LLMs, outperforming commonly used prompt engineering techniques.
Authors: Akbar Anbar Jafari, Cagri Ozcinar, Gholamreza Anbarjafari
Abstract: Contemporary machine learning models, including large language models, exhibit remarkable capabilities in static tasks yet falter in non-stationary environments due to rigid architectures that hinder continual adaptation and lifelong learning. Building upon the nested learning paradigm, which decomposes models into multi-level optimization problems with fixed update frequencies, this work proposes dynamic nested hierarchies as the next evolutionary step in advancing artificial intelligence and machine learning. Dynamic nested hierarchies empower models to autonomously adjust the number of optimization levels, their nesting structures, and update frequencies during training or inference, inspired by neuroplasticity to enable self-evolution without predefined constraints. This innovation addresses the anterograde amnesia in existing models, facilitating true lifelong learning by dynamically compressing context flows and adapting to distribution shifts. Through rigorous mathematical formulations, theoretical proofs of convergence, expressivity bounds, and sublinear regret in varying regimes, alongside empirical demonstrations of superior performance in language modeling, continual learning, and long-context reasoning, dynamic nested hierarchies establish a foundational advancement toward adaptive, general-purpose intelligence.
Authors: Yifeng Ding, Hung Le, Songyang Han, Kangrui Ruan, Zhenghui Jin, Varun Kumar, Zijian Wang, Anoop Deoras
Abstract: Training Large Language Models (LLMs) for multi-turn Tool-Integrated Reasoning (TIR) - where models iteratively reason, generate code, and verify through execution - remains challenging for existing reinforcement learning (RL) approaches. Current RL methods, exemplified by Group Relative Policy Optimization (GRPO), suffer from coarse-grained, trajectory-level rewards that provide insufficient learning signals for complex multi-turn interactions, leading to training stagnation. To address this issue, we propose Group Turn Policy Optimization (GTPO), a novel RL algorithm specifically designed for training LLMs on multi-turn TIR tasks. GTPO introduces three key innovations: (1) turn-level reward assignment that provides fine-grained feedback for individual turns, (2) return-based advantage estimation where normalized discounted returns are calculated as advantages, and (3) self-supervised reward shaping that exploits self-supervision signals from generated code to densify sparse binary outcome-based rewards. Our comprehensive evaluation demonstrates that GTPO outperforms GRPO by 3.0% on average across diverse reasoning benchmarks, establishing its effectiveness for advancing complex mathematical reasoning in the real world.
Authors: Dwipam Katariya, Snehita Varma, Akshat Shreemali, Benjamin Wu, Kalanand Mishra, Pranab Mohanty
Abstract: Transformer-based architectures are widely adopted in sequential recommendation systems, yet their application in Financial Services (FS) presents distinct practical and modeling challenges for real-time recommendation. These include:a) long-range user interactions (implicit and explicit) spanning both digital and physical channels generating temporally heterogeneous context, b) the presence of multiple interrelated products require coordinated models to support varied ad placements and personalized feeds, while balancing competing business goals. We propose FinTRec, a transformer-based framework that addresses these challenges and its operational objectives in FS. While tree-based models have traditionally been preferred in FS due to their explainability and alignment with regulatory requirements, our study demonstrate that FinTRec offers a viable and effective shift toward transformer-based architectures. Through historic simulation and live A/B test correlations, we show FinTRec consistently outperforms the production-grade tree-based baseline. The unified architecture, when fine-tuned for product adaptation, enables cross-product signal sharing, reduces training cost and technical debt, while improving offline performance across all products. To our knowledge, this is the first comprehensive study of unified sequential recommendation modeling in FS that addresses both technical and business considerations.
Authors: Nathan M. Roberts II, Xiaosong Du
Abstract: The rapid advancement of electric vertical take-off and landing (eVTOL) aircraft offers a promising opportunity to alleviate urban traffic congestion. Thus, developing optimal takeoff trajectories for minimum energy consumption becomes essential for broader eVTOL aircraft applications. Conventional optimal control methods (such as dynamic programming and linear quadratic regulator) provide highly efficient and well-established solutions but are limited by problem dimensionality and complexity. Deep reinforcement learning (DRL) emerges as a special type of artificial intelligence tackling complex, nonlinear systems; however, the training difficulty is a key bottleneck that limits DRL applications. To address these challenges, we propose the transformer-guided DRL to alleviate the training difficulty by exploring a realistic state space at each time step using a transformer. The proposed transformer-guided DRL was demonstrated on an optimal takeoff trajectory design of an eVTOL drone for minimal energy consumption while meeting takeoff conditions (i.e., minimum vertical displacement and minimum horizontal velocity) by varying control variables (i.e., power and wing angle to the vertical). Results presented that the transformer-guided DRL agent learned to take off with $4.57\times10^6$ time steps, representing 25% of the $19.79\times10^6$ time steps needed by a vanilla DRL agent. In addition, the transformer-guided DRL achieved 97.2% accuracy on the optimal energy consumption compared against the simulation-based optimal reference while the vanilla DRL achieved 96.3% accuracy. Therefore, the proposed transformer-guided DRL outperformed vanilla DRL in terms of both training efficiency as well as optimal design verification.
Authors: Grace Kim, Filip Svoboda, Nicholas Lane
Abstract: As Low Earth Orbit (LEO) satellite constellations rapidly expand to hundreds and thousands of spacecraft, the need for distributed on-board machine learning becomes critical to address downlink bandwidth limitations. Federated learning (FL) offers a promising framework to conduct collaborative model training across satellite networks. Realizing its benefits in space naturally requires addressing space-specific constraints, from intermittent connectivity to dynamics imposed by orbital motion. This work presents the first systematic feasibility analysis of adapting off-the-shelf FL algorithms for satellite constellation deployment. We introduce a comprehensive "space-ification" framework that adapts terrestrial algorithms (FedAvg, FedProx, FedBuff) to operate under orbital constraints, producing an orbital-ready suite of FL algorithms. We then evaluate these space-ified methods through extensive parameter sweeps across 768 constellation configurations that vary cluster sizes (1-10), satellites per cluster (1-10), and ground station networks (1-13). Our analysis demonstrates that space-adapted FL algorithms efficiently scale to constellations of up to 100 satellites, achieving performance close to the centralized ideal. Multi-month training cycles can be reduced to days, corresponding to a 9x speedup through orbital scheduling and local coordination within satellite clusters. These results provide actionable insights for future mission designers, enabling distributed on-board learning for more autonomous, resilient, and data-driven satellite operations.
Authors: Ruixin Zhang, Jon Donnelly, Zhicheng Guo, Ghazal Khalighinejad, Haiyang Huang, Alina Jade Barnett, Cynthia Rudin
Abstract: Large language models (LLMs) have exhibited remarkable capabilities across various domains. The ability to call external tools further expands their capability to handle real-world tasks. However, LLMs often follow an opaque reasoning process, which limits their usefulness in high-stakes domains where solutions need to be trustworthy to end users. LLMs can choose solutions that are unreliable and difficult to troubleshoot, even if better options are available. We address this issue by forcing LLMs to use external -- more reliable -- tools to solve problems when possible. We present a framework built on the tool-calling capabilities of existing LLMs to enable them to select the most reliable and easy-to-troubleshoot solution path, which may involve multiple sequential tool calls. We refer to this framework as LIT (LLMs with Inspectable Tools). In order to support LIT, we introduce a new and challenging benchmark dataset of 1,300 questions and a customizable set of reliability cost functions associated with a collection of specialized tools. These cost functions summarize how reliable each tool is and how easy it is to troubleshoot. For instance, a calculator is reliable across domains, whereas a linear prediction model is not reliable if there is distribution shift, but it is easy to troubleshoot. A tool that constructs a random forest is neither reliable nor easy to troubleshoot. These tools interact with the Harvard USPTO Patent Dataset and a new dataset of NeurIPS 2023 papers to solve mathematical, coding, and modeling problems of varying difficulty levels. We demonstrate that LLMs can achieve more reliable and informed problem-solving while maintaining task performance using our framework.
Authors: Zhengyang Shen, Hua Tu, Mayue Shi
Abstract: Neural networks exhibit severe brittleness to semantically irrelevant transformations. A mere 75ms electrocardiogram (ECG) phase shift degrades latent cosine similarity from 1.0 to 0.2, while sensor rotations collapse activity recognition performance with inertial measurement units (IMUs). We identify the root cause as "laissez-faire" representation learning, where latent spaces evolve unconstrained provided task performance is satisfied. We propose Structured Contrastive Learning (SCL), a framework that partitions latent space representations into three semantic groups: invariant features that remain consistent under given transformations (e.g., phase shifts or rotations), variant features that actively differentiate transformations via a novel variant mechanism, and free features that preserve task flexibility. This creates controllable push-pull dynamics where different latent dimensions serve distinct, interpretable purposes. The variant mechanism enhances contrastive learning by encouraging variant features to differentiate within positive pairs, enabling simultaneous robustness and interpretability. Our approach requires no architectural modifications and integrates seamlessly into existing training pipelines. Experiments on ECG phase invariance and IMU rotation robustness demonstrate superior performance: ECG similarity improves from 0.25 to 0.91 under phase shifts, while WISDM activity recognition achieves 86.65% accuracy with 95.38% rotation consistency, consistently outperforming traditional data augmentation. This work represents a paradigm shift from reactive data augmentation to proactive structural learning, enabling interpretable latent representations in neural networks.
Authors: Pranay Kumar Peddi, Dhrubajyoti Ghosh
Abstract: Deep graph learning has advanced Alzheimer's (AD) disease classification from MRI, but most models remain correlational, confounding demographic and genetic factors with disease specific features. We present Causal-GCN, an interventional graph convolutional framework that integrates do-calculus-based back-door adjustment to identify brain regions exerting stable causal influence on AD progression. Each subject's MRI is represented as a structural connectome where nodes denote cortical and subcortical regions and edges encode anatomical connectivity. Confounders such as age, sec, and APOE4 genotype are summarized via principal components and included in the causal adjustment set. After training, interventions on individual regions are simulated by serving their incoming edges and altering node features to estimate average causal effects on disease probability. Applied to 484 subjects from the ADNI cohort, Causal-GCN achieves performance comparable to baseline GNNs while providing interpretable causal effect rankings that highlight posterior, cingulate, and insular hubs consistent with established AD neuropathology.
Authors: Mathieu Dufour, Andrew Duncan
Abstract: Large language models trained on clinical text risk exposing sensitive patient information, yet differential privacy (DP) methods often severely degrade the diagnostic accuracy needed for deployment. Despite rapid progress in DP optimisation and text generation, it remains unclear which privacy-preserving strategy actually works best for clinical language tasks. We present the first systematic head-to-head comparison of four training pipelines for automated diagnostic coding from hospital discharge summaries. All pipelines use identical 1B-parameter models and matched privacy budgets to predict ICD-9 codes. At moderate and relaxed privacy budgets ($\varepsilon \in \{4, 6\}$), knowledge distillation from DP-trained teachers outperforms both direct DP-SGD and DP-synthetic data training, recovering up to 63\% of the non-private performance whilst maintaining strong empirical privacy (membership-inference AUC $\approx$ 0.5). These findings expose large differences in the privacy-utility trade-off across architectures and identify knowledge distillation as the most practical route to privacy-preserving clinical NLP.
Authors: Artur A. Oliveira, Mateus Espadoto, Roberto M. Cesar Jr., Roberto Hirata Jr
Abstract: We introduce Graph Memory (GM), a structured non-parametric framework that augments embedding-based inference with a compact, relational memory over region-level prototypes. Rather than treating each training instance in isolation, GM summarizes the embedding space into prototype nodes annotated with reliability indicators and connected by edges that encode geometric and contextual relations. This design unifies instance retrieval, prototype-based reasoning, and graph-based label propagation within a single inductive model that supports both efficient inference and faithful explanation. Experiments on synthetic and real datasets including breast histopathology (IDC) show that GM achieves accuracy competitive with $k$NN and Label Spreading while offering substantially better calibration and smoother decision boundaries, all with an order of magnitude fewer samples. By explicitly modeling reliability and relational structure, GM provides a principled bridge between local evidence and global consistency in non-parametric learning.
Authors: Halil S. Kelebek, Linnea M. Wolniewicz, Michael D. Vergalla, Simone Mestici, Giacomo Acciarini, Bala Poduval, Olga Verkhoglyadova, Madhulika Guhathakurta, Thomas E. Berger, Frank Soboczenski, At{\i}l{\i}m G\"une\c{s} Baydin
Abstract: The ionosphere is a critical component of near-Earth space, shaping GNSS accuracy, high-frequency communications, and aviation operations. For these reasons, accurate forecasting and modeling of ionospheric variability has become increasingly relevant. To address this gap, we present IonCast, a suite of deep learning models that include a GraphCast-inspired model tailored for ionospheric dynamics. IonCast leverages spatiotemporal learning to forecast global Total Electron Content (TEC), integrating diverse physical drivers and observational datasets. Validating on held-out storm-time and quiet conditions highlights improved skill compared to persistence. By unifying heterogeneous data with scalable graph-based spatiotemporal learning, IonCast demonstrates how machine learning can augment physical understanding of ionospheric variability and advance operational space weather resilience.
Authors: Jeffrey Jiang, Kevin Hong, Emily Kuczynski, Gregory Pottie
Abstract: While intelligent tutoring systems (ITSs) can use information from past students to personalize instruction, each new student is unique. Moreover, the education problem is inherently difficult because the learning process is only partially observable. We therefore develop a dynamic, time-series environment to simulate a classroom setting, with student-teacher interventions - including tutoring sessions, lectures, and exams. In particular, we design the simulated environment to allow for varying levels of probing interventions that can gather more information. Then, we develop reinforcement learning ITSs that combine learning the individual state of students while pulling from population information through the use of probing interventions. These interventions can reduce the difficulty of student estimation, but also introduce a cost-benefit decision to find a balance between probing enough to get accurate estimates and probing so often that it becomes disruptive to the student. We compare the efficacy of standard RL algorithms with several greedy rules-based heuristic approaches to find that they provide different solutions, but with similar results. We also highlight the difficulty of the problem with increasing levels of hidden information, and the boost that we get if we allow for probing interventions. We show the flexibility of both heuristic and RL policies with regards to changing student population distributions, finding that both are flexible, but RL policies struggle to help harder classes. Finally, we test different course structures with non-probing policies and we find that our policies are able to boost the performance of quiz and midterm structures more than we can in a finals-only structure, highlighting the benefit of having additional information.
Authors: Zachariah Farahany, Jiawei Wu, K M Sajjadul Islam, Praveen Madiraju
Abstract: COVID-19 is a respiratory disease that caused a global pandemic in 2019. It is highly infectious and has the following symptoms: fever or chills, cough, shortness of breath, fatigue, muscle or body aches, headache, the new loss of taste or smell, sore throat, congestion or runny nose, nausea or vomiting, and diarrhea. These symptoms vary in severity; some people with many risk factors have been known to have lengthy hospital stays or die from the disease. In this paper, we analyze patients' electronic health records (EHR) to predict the severity of their COVID-19 infection using the length of stay (LOS) as our measurement of severity. This is an imbalanced classification problem, as many people have a shorter LOS rather than a longer one. To combat this problem, we synthetically create alternate oversampled training data sets. Once we have this oversampled data, we run it through an Artificial Neural Network (ANN), which during training has its hyperparameters tuned using Bayesian optimization. We select the model with the best F1 score and then evaluate it and discuss it.
Authors: Yun Kwan Kim
Abstract: Clinical decision support systems (CDSSs) have been widely utilized to support the decisions made by cardiologists when detecting and classifying arrhythmia from electrocardiograms. However, forming a CDSS for the arrhythmia classification task is challenging due to the varying lengths of arrhythmias. Although the onset time of arrhythmia varies, previously developed methods have not considered such conditions. Thus, we propose a framework that consists of (i) local and global extraction and (ii) local-global information fusion with attention to enable arrhythmia detection and classification within a constrained input length. The framework's performance was evaluated in terms of 10-class and 4-class arrhythmia detection, focusing on identifying the onset and ending point of arrhythmia episodes and their duration using the MIT-BIH arrhythmia database (MITDB) and the MIT-BIH atrial fibrillation database (AFDB). Duration, episode, and Dice score performances resulted in overall F1-scores of 96.45%, 82.05%, and 96.31% on the MITDB and 97.57%, 98.31%, and 97.45% on the AFDB, respectively. The results demonstrated statistically superior performance compared to those of the benchmark models. To assess the generalization capability of the proposed method, an MITDB-trained model and MIT-BIH malignant ventricular arrhythmia database-trained model were tested AFDB and MITDB, respectively. Superior performance was attained compared with that of a state-of-the-art model. The proposed method effectively captures both local and global information and dynamics without significant information loss. Consequently, arrhythmias can be detected with greater accuracy, and their occurrence times can be precisely determined, enabling the clinical field to develop more accurate treatment plans based on the proposed method.
Authors: Zisong Wang, Xuanyu Wang, Hang Chen, Haizhou Wang, Yuxin Chen, Yihang Xu, Yunhe Yuan, Lihuan Luo, Xitong Ling, Xiaoping Liu
Abstract: Precise prognostic stratification of colorectal cancer (CRC) remains a major clinical challenge due to its high heterogeneity. The conventional TNM staging system is inadequate for personalized medicine. We aimed to develop and validate a novel multiple instance learning model TDAM-CRC using histopathological whole-slide images for accurate prognostic prediction and to uncover its underlying molecular mechanisms. We trained the model on the TCGA discovery cohort (n=581), validated it in an independent external cohort (n=1031), and further we integrated multi-omics data to improve model interpretability and identify novel prognostic biomarkers. The results demonstrated that the TDAM-CRC achieved robust risk stratification in both cohorts. Its predictive performance significantly outperformed the conventional clinical staging system and multiple state-of-the-art models. The TDAM-CRC risk score was confirmed as an independent prognostic factor in multivariable analysis. Multi-omics analysis revealed that the high-risk subtype is closely associated with metabolic reprogramming and an immunosuppressive tumor microenvironment. Through interaction network analysis, we identified and validated Mitochondrial Ribosomal Protein L37 (MRPL37) as a key hub gene linking deep pathomic features to clinical prognosis. We found that high expression of MRPL37, driven by promoter hypomethylation, serves as an independent biomarker of favorable prognosis. Finally, we constructed a nomogram incorporating the TDAM-CRC risk score and clinical factors to provide a precise and interpretable clinical decision-making tool for CRC patients. Our AI-driven pathological model TDAM-CRC provides a robust tool for improved CRC risk stratification, reveals new molecular targets, and facilitates personalized clinical decision-making.
Authors: Xiancheng Wang, Lin Wang, Rui Wang, Zhibo Zhang, Minghang Zhao
Abstract: Time-series anomaly detection plays a critical role in numerous real-world applications, including industrial monitoring and fault diagnosis. Recently, Mamba-based state-space models have shown remarkable efficiency in long-sequence modeling. However, directly applying Mamba to anomaly detection tasks still faces challenges in capturing complex temporal patterns and nonlinear dynamics. In this paper, we propose Fourier-KAN-Mamba, a novel hybrid architecture that integrates Fourier layer, Kolmogorov-Arnold Networks (KAN), and Mamba selective state-space model. The Fourier layer extracts multi-scale frequency features, KAN enhances nonlinear representation capability, and a temporal gating control mechanism further improves the model's ability to distinguish normal and anomalous patterns. Extensive experiments on MSL, SMAP, and SWaT datasets demonstrate that our method significantly outperforms existing state-of-the-art approaches. Keywords: time-series anomaly detection, state-space model, Mamba, Fourier transform, Kolmogorov-Arnold Network
Authors: Wei-hsiang Yen, Lyn Chao-ling Chen
Abstract: The innovation of the study is that the deep learning method and sentiment analysis are integrated in traditional business model analysis and forecasting, and the research subject is TSMC for industry trend prediction of semiconductor industry in Taiwan. For the rapid market changes and development of wafer technologies of semiconductor industry, traditional data analysis methods not perform well in the high variety and time series data. Textual data and time series data were collected from seasonal reports of TSMC including financial information. Textual data through sentiment analysis by considering the event intervention both from internal events of the company and the external global events. Using the sentiment-enhanced time series data, the LSTM model was adopted for predicting industry trend of TSMC. The prediction results reveal significant development of wafer technology of TSMC and the potential threatens in the global market, and matches the product released news of TSMC and the international news. The contribution of the work performed accurately in industry trend prediction of the semiconductor industry by considering both the internal and external event intervention, and the prediction results provide valuable information of semiconductor industry both in research and business aspects.
Authors: Huifan Zhang, Pingqiang Zhou
Abstract: Conventional radio frequency (RF) passive components modeling based on machine learning requires extensive electromagnetic (EM) simulations to cover geometric and frequency design spaces, creating computational bottlenecks. In this paper, we introduce an uncertainty-aware Bayesian online learning framework for efficient parametric modeling of RF passive components, which includes: 1) a Bayesian neural network with reconfigurable heads for joint geometric-frequency domain modeling while quantifying uncertainty; 2) an adaptive sampling strategy that simultaneously optimizes training data sampling across geometric parameters and frequency domain using uncertainty guidance. Validated on three RF passive components, the framework achieves accurate modeling while using only 2.86% EM simulation time compared to traditional ML-based flow, achieving a 35 times speedup.
Authors: Andrew Amos, Joanne Lee, Tarun Sen Gupta, Bunmi S. Malau-Aduli
Abstract: Past efforts to map the Medline database have been limited to small subsets of the available data because of the exponentially increasing memory and processing demands of existing algorithms. We designed a novel algorithm for sparse matrix multiplication that allowed us to apply a self-organizing map to the entire Medline dataset, allowing for a more complete map of existing medical knowledge. The algorithm also increases the feasibility of refining the self-organizing map to account for changes in the dataset over time.
Authors: Xiaoxuan Wang, Bo Liu, Song Jiang, Jingzhou Liu, Jingyuan Qi, Xia Chen, Baosheng He
Abstract: The reasoning capabilities of large language models (LLMs) have been significantly improved through reinforcement learning (RL). Nevertheless, LLMs still struggle to consistently verify their own reasoning traces. This raises the research question of how to enhance the self-verification ability of LLMs and whether such an ability can further improve reasoning performance. In this work, we propose GRPO-Verif, an algorithm that jointly optimizes solution generation and self-verification within a unified loss function, with an adjustable hyperparameter controlling the weight of the verification signal. Experimental results demonstrate that our method enhances self-verification capability while maintaining comparable performance in reasoning.
Authors: Hyo-Jeong Jang, Hye-Bin Shin, Kang Yin
Abstract: Deep learning models perform best with abundant, high-quality labels, yet such conditions are rarely achievable in EEG-based emotion recognition. Electroencephalogram (EEG) signals are easily corrupted by artifacts and individual variability, while emotional labels often stem from subjective and inconsistent reports-making robust affective decoding particularly difficult. We propose an uncertainty-aware active learning framework that enhances robustness to label noise by jointly leveraging model uncertainty and cross-modal consistency. Instead of relying solely on EEG-based uncertainty estimates, the method evaluates cross-modal alignment to determine whether uncertainty originates from cognitive ambiguity or sensor noise. A representation alignment module embeds EEG and face features into a shared latent space, enforcing semantic coherence between modalities. Residual discrepancies are treated as noise-induced inconsistencies, and these samples are selectively queried for oracle feedback during active learning. This feedback-driven process guides the network toward reliable, informative samples and reduces the impact of noisy labels. Experiments on the ASCERTAIN dataset examine the efficiency and robustness of ours, highlighting its potential as a data-efficient and noise-tolerant approach for EEG-based affective decoding in brain-computer interface systems.
Authors: Andrew Gracyk
Abstract: It has been discovered that latent-Euclidean variational autoencoders (VAEs) admit, in various capacities, Riemannian structure. We adapt these arguments but for complex VAEs with a complex latent stage. We show that complex VAEs reveal to some level K\"ahler geometric structure. Our methods will be tailored for decoder geometry. We derive the Fisher information metric in the complex case under a latent complex Gaussian regularization with trivial relation matrix. It is well known from statistical information theory that the Fisher information coincides with the Hessian of the Kullback-Leibler (KL) divergence. Thus, the metric K\"ahler potential relation is exactly achieved under relative entropy. We propose a K\"ahler potential derivative of complex Gaussian mixtures that has rough equivalence to the Fisher information metric while still being faithful to the underlying K\"ahler geometry. Computation of the metric via this potential is efficient, and through our potential, valid as a plurisubharmonic (PSH) function, large scale computational burden of automatic differentiation is displaced to small scale. We show that we can regularize the latent space with decoder geometry, and that we can sample in accordance with a weighted complex volume element. We demonstrate these strategies, at the exchange of sample variation, yield consistently smoother representations and fewer semantic outliers.
Authors: Yi Xu, Zhigang Chen, Rui Wang, Yangfan Li, Fengxiao Tang, Ming Zhao, Jiaqi Liu
Abstract: In industrial equipment monitoring, fault diagnosis is critical for ensuring system reliability and enabling predictive maintenance. However, the scarcity of fault data, due to the rarity of fault events and the high cost of data annotation, significantly hinders data-driven approaches. Existing time-series generation models, optimized for abundant normal data, struggle to capture fault distributions in few-shot scenarios, producing samples that lack authenticity and diversity due to the large domain gap and high intra-class variability of faults. To address this, we propose a novel few-shot fault time-series generation framework based on diffusion models. Our approach employs a positive-negative difference adapter, leveraging pre-trained normal data distributions to model the discrepancies between normal and fault domains for accurate fault synthesis. Additionally, a diversity loss is introduced to prevent mode collapse, encouraging the generation of diverse fault samples through inter-sample difference regularization. Experimental results demonstrate that our model significantly outperforms traditional methods in authenticity and diversity, achieving state-of-the-art performance on key benchmarks.
Authors: Le Tung Giang, Vu Hoang Viet, Nguyen Xuan Tung, Trinh Van Chien, Won-Joo Hwang
Abstract: The vehicle routing problem (VRP) is a fundamental NP-hard task in intelligent transportation systems with broad applications in logistics and distribution. Deep reinforcement learning (DRL) with Graph Neural Networks (GNNs) has shown promise, yet classical models rely on large multi-layer perceptrons (MLPs) that are parameter-heavy and memory-bound. We propose a Quantum Graph Attention Network (Q-GAT) within a DRL framework, where parameterized quantum circuits (PQCs) replace conventional MLPs at critical readout stages. The hybrid model maintains the expressive capacity of graph attention encoders while reducing trainable parameters by more than 50%. Using proximal policy optimization (PPO) with greedy and stochastic decoding, experiments on VRP benchmarks show that Q-GAT achieves faster convergence and reduces routing cost by about 5% compared with classical GAT baselines. These results demonstrate the potential of PQC-enhanced GNNs as compact and effective solvers for large-scale routing and logistics optimization.
Authors: Yuxuan Gu, Weimin Bai, Yifei Wang, Weijian Luo, He Sun
Abstract: Masked auto-regressive diffusion models (MAR) benefit from the expressive modeling ability of diffusion models and the flexibility of masked auto-regressive ordering. However, vanilla MAR suffers from slow inference due to its hierarchical inference mechanism: an outer AR unmasking loop and an inner diffusion denoising chain. Such decoupled structure not only harm the generation efficiency but also hinder the practical use of MAR for reinforcement learning (RL), an increasingly critical paradigm for generative model post-training.To address this fundamental issue, we introduce MARVAL (Masked Auto-regressive Variational Acceleration), a distillation-based framework that compresses the diffusion chain into a single AR generation step while preserving the flexible auto-regressive unmasking order. Such a distillation with MARVAL not only yields substantial inference acceleration but, crucially, makes RL post-training with verifiable rewards practical, resulting in scalable yet human-preferred fast generative models. Our contributions are twofold: (1) a novel score-based variational objective for distilling masked auto-regressive diffusion models into a single generation step without sacrificing sample quality; and (2) an efficient RL framework for masked auto-regressive models via MARVAL-RL. On ImageNet 256*256, MARVAL-Huge achieves an FID of 2.00 with more than 30 times speedup compared with MAR-diffusion, and MARVAL-RL yields consistent improvements in CLIP and image-reward scores on ImageNet datasets with entity names. In conclusion, MARVAL demonstrates the first practical path to distillation and RL of masked auto-regressive diffusion models, enabling fast sampling and better preference alignments.
Authors: Ranfei Chen, Ming Chen, Kaifei Wang
Abstract: Diffusion Large Language Models (dLLMs) are rapidly emerging alongside autoregressive models as a powerful paradigm for complex reasoning, with reinforcement learning increasingly used for downstream alignment. Existing trajectory-based RL methods uniformly allocate policy gradients across denoising steps, implicitly treating all steps as equally important. We challenge this assumption by analyzing trajectories with several step-level metrics: entropy-based uncertainty, Confidence-Margin (CM) uncertainty, and Rate of Entropy Change (RoEC). These reveal structured "zones of confusion": transient spikes in uncertainty and instability that strongly predict final success or failure, while most steps remain stable. We propose Adaptive Trajectory Policy Optimization (ATPO), a lightweight step-selection strategy that dynamically reallocates gradient updates to these high-leverage steps without changing the RL objective, rewards, or compute budget. Using a hybrid RoEC+CM rule, ATPO delivers substantial gains in reasoning accuracy and training stability across benchmarks, showing that exploiting trajectory dynamics is key to advancing dLLM RL.
Authors: Tung Giang Le, Xuan Tung Nguyen, Won-Joo Hwang
Abstract: Increasing wireless network complexity demands scalable resource management. Classical GNNs excel at graph learning but incur high computational costs in large-scale settings. We present a fully quantum Graph Neural Network (QGNN) that implements message passing via Parameterized Quantum Circuits (PQCs). Our Quantum Graph Convolutional Layers (QGCLs) encode features into quantum states, process graphs with NISQ-compatible unitaries, and retrieve embeddings through measurement. Applied to D2D power control for SINR maximization, our QGNN matches classical performance with fewer parameters and inherent parallelism. This end-to-end PQC-based GNN marks a step toward quantum-accelerated wireless optimization.
Authors: Kai Yang, Xin Xu, Yangkun Chen, Weijie Liu, Jiafei Lyu, Zichuan Lin, Deheng Ye, Saiyong Yang
Abstract: Long-term training of large language models (LLMs) requires maintaining stable exploration to prevent the model from collapsing into sub-optimal behaviors. Entropy is crucial in this context, as it controls exploration and helps avoid premature convergence to sub-optimal solutions. However, existing reinforcement learning methods struggle to maintain an appropriate level of entropy, as the training process involves a mix of positive and negative samples, each affecting entropy in different ways across steps. To address this, we propose Entropy stablilization via Proportional-Integral Control (EntroPIC), a novel method that adaptively adjusts the influence of positive and negative samples by dynamically tuning their loss coefficients. This approach stabilizes entropy throughout training, ensuring efficient exploration and steady progress. We provide a comprehensive theoretical analysis for both on-policy and off-policy learning settings, demonstrating that EntroPIC is effective at controlling entropy in large-scale LLM training. Experimental results show that our method successfully maintains desired entropy levels, enabling stable and optimal RL training for LLMs.
Authors: Jin Ye, Lingmei Wang, Shujian Zhang, Haihang WU
Abstract: With the global energy transition and rapid development of renewable energy, the scheduling optimization challenge for combined power-heat systems under new energy integration and multiple uncertainties has become increasingly prominent. Addressing this challenge, this study proposes an intelligent scheduling method based on the improved Dual-Delay Deep Deterministic Policy Gradient (PVTD3) algorithm. System optimization is achieved by introducing a penalty term for grid power purchase variations. Simulation results demonstrate that under three typical scenarios (10%, 20%, and 30% renewable penetration), the PVTD3 algorithm reduces the system's comprehensive cost by 6.93%, 12.68%, and 13.59% respectively compared to the traditional TD3 algorithm. Concurrently, it reduces the average fluctuation amplitude of grid power purchases by 12.8%. Regarding energy storage management, the PVTD3 algorithm reduces the end-time state values of low-temperature thermal storage tanks by 7.67-17.67 units while maintaining high-temperature tanks within the 3.59-4.25 safety operating range. Multi-scenario comparative validation demonstrates that the proposed algorithm not only excels in economic efficiency and grid stability but also exhibits superior sustainable scheduling capabilities in energy storage device management.
Authors: Chengze Du, Heng Xu, Zhiwei Yu, Bo Liu, Jialong Li
Abstract: Network tomography aims to infer hidden network states, such as link performance, traffic load, and topology, from external observations. Most existing methods solve these problems separately and depend on limited task-specific signals, which limits generalization and interpretability. We present PLATONT, a unified framework that models different network indicators (e.g., delay, loss, bandwidth) as projections of a shared latent network state. Guided by the Platonic Representation Hypothesis, PLATONT learns this latent state through multimodal alignment and contrastive learning. By training multiple tomography tasks within a shared latent space, it builds compact and structured representations that improve cross-task generalization. Experiments on synthetic and real-world datasets show that PLATONT consistently outperforms existing methods in link estimation, topology inference, and traffic prediction, achieving higher accuracy and stronger robustness under varying network conditions.
Authors: Yanchen Xu, Ziheng Jiao, Hongyuan Zhang, Xuelong Li
Abstract: The Group Relative Policy Optimization (GRPO), a reinforcement learning method used to fine-tune large language models (LLMs), has proved its effectiveness in practical applications such as DeepSeek-R1. It raises a question whether GRPO can be generalized to representation learning models. In this paper, we propose Group Relative Policy Optimization for Representation Model (GRPO-RM), and investigate the performance of GRPO-like policy in post-training representation models. Specifically, our method establishes a predefined output set to functionally replace token sequence sampling in LLMs, thereby generating an output group, which is essential for the probability-driven optimization of GRPO. In addition, a specialized reward function is designed to accommodate the properties of representation models. Extensive experiments are conducted on various real-world datasets to validate the effectiveness of our proposed method.
Authors: Hyeongheon Cha, Dong Min Kim, Hye Won Chung, Taesik Gong, Sung-Ju Lee
Abstract: Test-Time Adaptation (TTA) adjusts models using unlabeled test data to handle dynamic distribution shifts. However, existing methods rely on frequent adaptation and high computational cost, making them unsuitable for resource-constrained edge environments. To address this, we propose SNAP, a sparse TTA framework that reduces adaptation frequency and data usage while preserving accuracy. SNAP maintains competitive accuracy even when adapting based on only 1% of the incoming data stream, demonstrating its robustness under infrequent updates. Our method introduces two key components: (i) Class and Domain Representative Memory (CnDRM), which identifies and stores a small set of samples that are representative of both class and domain characteristics to support efficient adaptation with limited data; and (ii) Inference-only Batch-aware Memory Normalization (IoBMN), which dynamically adjusts normalization statistics at inference time by leveraging these representative samples, enabling efficient alignment to shifting target domains. Integrated with five state-of-the-art TTA algorithms, SNAP reduces latency by up to 93.12%, while keeping the accuracy drop below 3.3%, even across adaptation rates ranging from 1% to 50%. This demonstrates its strong potential for practical use on edge devices serving latency-sensitive applications. The source code is available at https://github.com/chahh9808/SNAP.
Authors: Rayen Dhahri, Steffen Urban
Abstract: Specialized edge accelerators rely on low-bit quantization, but vendor compilers differ in scaling, clipping, and kernel support, often as black boxes. The same floating-point (FP) checkpoint can therefore yield inconsistent accuracy across backends, forcing practitioners to tweak flags or refactor models to vendor-friendly operator subsets. We introduce Quant-Trim, a training-phase method that produces a hardware-neutral checkpoint robust to backend and precision choices. It combines progressive fake quantization to align training with the deployed integer grid and reverse pruning to tame outlier-driven scale inflation while preserving learnability. Quant-Trim is agnostic to quantization schemes (symmetric/asymmetric,per-tensor/per-channel, INT8/INT4) and requires no vendor-specific graph changes.Across models and tasks, it narrows the FP,low-bit gap, reduces dependence on compiler heuristics/calibration, and avoids per-backend retraining. We report accuracy and edge metrics latency, throughput, energy/inference, and cost under static/dynamic activation scaling and varying operator coverage.
Authors: Atharva Pandey, Abhilash Neog, Gautam Jajoo
Abstract: Time-series Foundation Models (TSFMs) have recently emerged as a universal paradigm for learning across diverse temporal domains. However, despite their empirical success, the internal mechanisms by which these models represent fundamental time-series concepts remain poorly understood. In this work, we undertake a systematic investigation of concept interpretability in TSFMs. Specifically, we examine: (i) which layers encode which concepts, (ii) whether concept parameters are linearly recoverable, (iii) how representations evolve in terms of concept disentanglement and abstraction across model depth, and (iv) how models process compositions of concepts. We systematically probe these questions using layer-wise analyses, linear recoverability tests, and representation similarity measures, providing a structured account of TSFM semantics. The resulting insights show that early layers mainly capture local, time-domain patterns (e.g., AR(1), level shifts, trends), while deeper layers encode dispersion and change-time signals, with spectral and warping factors remaining the hardest to recover linearly. In compositional settings, however, probe performance degrades, revealing interference between concepts. This highlights that while atomic concepts are reliably localized, composition remains a challenge, underscoring a key limitation in current TSFMs' ability to represent interacting temporal phenomena.
Authors: Huseyin Goksu
Abstract: Spectral Graph Neural Networks (GNNs) based on polynomial filters, such as ChebyNet, suffer from two critical limitations: 1) performance collapse on "heterophilic" graphs and 2) performance collapse at high polynomial degrees (K), known as over-smoothing. Both issues stem from the static, low-pass nature of standard filters. In this work, we propose `KrawtchoukNet`, a GNN filter based on the discrete Krawtchouk polynomials. We demonstrate that `KrawtchoukNet` provides a unified solution to both problems through two key design choices. First, by fixing the polynomial's domain N to a small constant (e.g., N=20), we create the first GNN filter whose recurrence coefficients are \textit{inherently bounded}, making it exceptionally robust to over-smoothing (achieving SOTA results at K=10). Second, by making the filter's shape parameter p learnable, the filter adapts its spectral response to the graph data. We show this adaptive nature allows `KrawtchoukNet` to achieve SOTA performance on challenging heterophilic benchmarks (Texas, Cornell), decisively outperforming standard GNNs like GAT and APPNP.
Authors: Huseyin Goksu
Abstract: Spectral Graph Neural Networks (GNNs) suffer from two critical limitations: poor performance on "heterophilic" graphs and performance collapse at high polynomial degrees (K), known as over-smoothing. Both issues stem from the static, low-pass nature of standard filters (e.g., ChebyNet). While adaptive polynomial filters, such as the discrete MeixnerNet, have emerged as a potential unified solution, their extension to the continuous domain and stability with unbounded coefficients remain open questions. In this work, we propose `LaguerreNet`, a novel GNN filter based on continuous Laguerre polynomials. `LaguerreNet` learns the filter's spectral shape by making its core alpha parameter trainable, thereby advancing the adaptive polynomial approach. We solve the severe O(k^2) numerical instability of these unbounded polynomials using a `LayerNorm`-based stabilization technique. We demonstrate experimentally that this approach is highly effective: 1) `LaguerreNet` achieves state-of-the-art results on challenging heterophilic benchmarks. 2) It is exceptionally robust to over-smoothing, with performance peaking at K=10, an order of magnitude beyond where ChebyNet collapses.
Authors: Kadir-Kaan \"Ozer, Ren\'e Ebeling, Markus Enzweiler
Abstract: Automotive telemetry data exhibits slow drifts and fast spikes, often within the same sequence, making reliable anomaly detection challenging. Standard reconstruction-based methods, including sequence variational autoencoders (VAEs), use a single latent process and therefore mix heterogeneous time scales, which can smooth out spikes or inflate variances and weaken anomaly separation. In this paper, we present STREAM-VAE, a variational autoencoder for anomaly detection in automotive telemetry time-series data. Our model uses a dual-path encoder to separate slow drift and fast spike signal dynamics, and a decoder that represents transient deviations separately from the normal operating pattern. STREAM-VAE is designed for deployment, producing stable anomaly scores across operating modes for both in-vehicle monitors and backend fleet analytics. Experiments on an automotive telemetry dataset and the public SMD benchmark show that explicitly separating drift and spike dynamics improves robustness compared to strong forecasting, attention, graph, and VAE baselines.
Authors: Nathanael Bosch, Oleksandr Shchur, Nick Erickson, Michael Bohlke-Schneider, Caner T\"urkmen
Abstract: Ensembling is a powerful technique for improving the accuracy of machine learning models, with methods like stacking achieving strong results in tabular tasks. In time series forecasting, however, ensemble methods remain underutilized, with simple linear combinations still considered state-of-the-art. In this paper, we systematically explore ensembling strategies for time series forecasting. We evaluate 33 ensemble models -- both existing and novel -- across 50 real-world datasets. Our results show that stacking consistently improves accuracy, though no single stacker performs best across all tasks. To address this, we propose a multi-layer stacking framework for time series forecasting, an approach that combines the strengths of different stacker models. We demonstrate that this method consistently provides superior accuracy across diverse forecasting scenarios. Our findings highlight the potential of stacking-based methods to improve AutoML systems for time series forecasting.
Authors: Yinan Yu, Falk Dippel, Christina E. Lundberg, Martin Lindgren, Annika Rosengren, Martin Adiels, Helen Sj\"oland
Abstract: Objective: Machine learning (ML) predictive models are often developed without considering downstream value trade-offs and clinical interpretability. This paper introduces a cost-aware prediction (CAP) framework that combines cost-benefit analysis assisted by large language model (LLM) agents to communicate the trade-offs involved in applying ML predictions. Materials and Methods: We developed an ML model predicting 1-year mortality in patients with heart failure (N = 30,021, 22% mortality) to identify those eligible for home care. We then introduced clinical impact projection (CIP) curves to visualize important cost dimensions - quality of life and healthcare provider expenses, further divided into treatment and error costs, to assess the clinical consequences of predictions. Finally, we used four LLM agents to generate patient-specific descriptions. The system was evaluated by clinicians for its decision support value. Results: The eXtreme gradient boosting (XGB) model achieved the best performance, with an area under the receiver operating characteristic curve (AUROC) of 0.804 (95% confidence interval (CI) 0.792-0.816), area under the precision-recall curve (AUPRC) of 0.529 (95% CI 0.502-0.558) and a Brier score of 0.135 (95% CI 0.130-0.140). Discussion: The CIP cost curves provided a population-level overview of cost composition across decision thresholds, whereas LLM-generated cost-benefit analysis at individual patient-levels. The system was well received according to the evaluation by clinicians. However, feedback emphasizes the need to strengthen the technical accuracy for speculative tasks. Conclusion: CAP utilizes LLM agents to integrate ML classifier outcomes and cost-benefit analysis for more transparent and interpretable decision support.
Authors: Eddie Conti, \'Alvaro Parafita, Axel Brando
Abstract: Assessing the importance of individual features in Machine Learning is critical to understand the model's decision-making process. While numerous methods exist, the lack of a definitive ground truth for comparison highlights the need for alternative, well-founded measures. This paper introduces a novel post-hoc local feature importance method called Counterfactual Importance Distribution (CID). We generate two sets of positive and negative counterfactuals, model their distributions using Kernel Density Estimation, and rank features based on a distributional dissimilarity measure. This measure, grounded in a rigorous mathematical framework, satisfies key properties required to function as a valid metric. We showcase the effectiveness of our method by comparing with well-established local feature importance explainers. Our method not only offers complementary perspectives to existing approaches, but also improves performance on faithfulness metrics (both for comprehensiveness and sufficiency), resulting in more faithful explanations of the system. These results highlight its potential as a valuable tool for model analysis.
Authors: Lingxiang Wang, Hainan Zhang, Zhiming Zheng
Abstract: Domain-specific post-training often causes catastrophic forgetting, making foundation models lose their general reasoning ability and limiting their adaptability to dynamic real-world environments. Preserving general capabilities while acquiring downstream domain knowledge is a central challenge for large language and multimodal models. Traditional continual learning methods, such as regularization, replay and architectural isolation, suffer from poor downstream performance, reliance on inaccessible historical data, or additional parameter overhead. While recent parameter-efficient tuning (PET) methods can alleviate forgetting, their effectiveness strongly depends on the choice of parameters and update strategies. In this paper, we introduce PIECE, a Parameter Importance Estimation-based Continual Enhancement method that preserves general ability while efficiently learning domain knowledge without accessing prior training data or increasing model parameters. PIECE selectively updates only 0.1% of core parameters most relevant to new tasks, guided by two importance estimators: PIECE-F based on Fisher Information, and PIECE-S based on a second-order normalization that combines gradient and curvature information. Experiments across three language models and two multimodal models show that PIECE maintains general capabilities and achieves state-of-the-art continual learning performance across diverse downstream tasks. Our results highlight a practical path to scalable, domain-adaptive foundation models without catastrophic forgetting.
Authors: Kunyu Zhang, Mingxuan Wang, Xiangjie Shi, Haoxing Xu, Chao Zhang
Abstract: The brain age is a key indicator of brain health. While electroencephalography (EEG) is a practical tool for this task, existing models struggle with the common challenge of imperfect medical data, such as learning a ``normal'' baseline from weakly supervised, healthy-only cohorts. This is a critical anomaly detection task for identifying disease, but standard models are often black boxes lacking an interpretable structure. We propose EVA-Net, a novel framework that recasts brain age as an interpretable anomaly detection problem. EVA-Net uses an efficient, sparsified-attention Transformer to model long EEG sequences. To handle noise and variability in imperfect data, it employs a Variational Information Bottleneck to learn a robust, compressed representation. For interpretability, this representation is aligned to a continuous prototype network that explicitly learns the normative healthy aging manifold. Trained on 1297 healthy subjects, EVA-Net achieves state-of-the-art accuracy. We validated its anomaly detection capabilities on an unseen cohort of 27 MCI and AD patients. This pathological group showed significantly higher brain-age gaps and a novel Prototype Alignment Error, confirming their deviation from the healthy manifold. EVA-Net provides an interpretable framework for healthcare intelligence using imperfect medical data.
Authors: Hany Abdulsamad, \'Angel F. Garc\'ia-Fern\'andez, Simo S\"arkk\"a
Abstract: We present a class of algorithms for state estimation in nonlinear, non-Gaussian state-space models. Our approach is based on a variational Lagrangian formulation that casts Bayesian inference as a sequence of entropic trust-region updates subject to dynamic constraints. This framework gives rise to a family of forward-backward algorithms, whose structure is determined by the chosen factorization of the variational posterior. By focusing on Gauss--Markov approximations, we derive recursive schemes with favorable computational complexity. For general nonlinear, non-Gaussian models we close the recursions using generalized statistical linear regression and Fourier--Hermite moment matching.
Authors: Amir Rezaei Balef, Mykhailo Koshil, Katharina Eggensperger
Abstract: Despite the architectural similarities between tabular in-context learning (ICL) models and large language models (LLMs), little is known about how individual layers contribute to tabular prediction. In this paper, we investigate how the latent spaces evolve across layers in tabular ICL models, identify potential redundant layers, and compare these dynamics with those observed in LLMs. We analyze TabPFN and TabICL through the "layers as painters" perspective, finding that only subsets of layers share a common representational language, suggesting structural redundancy and offering opportunities for model compression and improved interpretability.
Authors: Michel Tokic, Slobodan Djukanovi\'c, Anja von Beuningen, Cheng Feng
Abstract: This paper introduces a classification method using in-context learning in time-series foundation models (TSFM). We show how data, which was not part of the TSFM training data corpus, can be classified without the need of finetuning the model. Examples are represented in the form of targets (class id) and covariates (data matrix) within the prompt of the model, which enables to classify an unknown covariate data pattern alongside the forecast axis through in-context learning. We apply this method to vibration data for assessing the health state of a bearing within a servo-press motor. The method transforms frequency domain reference signals into pseudo time-series patterns, generates aligned covariate and target signals, and uses the TSFM to predict probabilities how classified data corresponds to predefined labels. Leveraging the scalability of pre-trained models this method demonstrates efficacy across varied operational conditions. This marks significant progress beyond custom narrow AI solutions towards broader, AI-driven maintenance systems.
Authors: Ouiame Marnissi, Hajar EL Hammouti, El Houcine Bergou
Abstract: Federated learning (FL) enables collaborative model training across distributed devices while preserving data privacy. However, balancing energy efficiency and fair participation while ensuring high model accuracy remains challenging in wireless edge systems due to heterogeneous resources, unequal client contributions, and limited communication capacity. To address these challenges, we propose FairEnergy, a fairness-aware energy minimization framework that integrates a contribution score capturing both the magnitude of updates and their compression ratio into the joint optimization of device selection, bandwidth allocation, and compression level. The resulting mixed-integer non-convex problem is solved by relaxing binary selection variables and applying Lagrangian decomposition to handle global bandwidth coupling, followed by per-device subproblem optimization. Experiments on non-IID data show that FairEnergy achieves higher accuracy while reducing energy consumption by up to 79\% compared to baseline strategies.
Authors: Chen Zhang, Wei Zuo, Bingyang Cheng, Yikun Wang, Wei-Bin Kou, Yik Chung WU, Ngai Wong
Abstract: Implicit Neural Representations (INRs) parameterize continuous signals via multilayer perceptrons (MLPs), enabling compact, resolution-independent modeling for tasks like image, audio, and 3D reconstruction. However, fitting high-resolution signals demands optimizing over millions of coordinates, incurring prohibitive computational costs. To address it, we propose NTK-Guided Implicit Neural Teaching (NINT), which accelerates training by dynamically selecting coordinates that maximize global functional updates. Leveraging the Neural Tangent Kernel (NTK), NINT scores examples by the norm of their NTK-augmented loss gradients, capturing both fitting errors and heterogeneous leverage (self-influence and cross-coordinate coupling). This dual consideration enables faster convergence compared to existing methods. Through extensive experiments, we demonstrate that NINT significantly reduces training time by nearly half while maintaining or improving representation quality, establishing state-of-the-art acceleration among recent sampling-based strategies.
Authors: Nika Haghtalab, Omar Montasser, Mingda Qiao
Abstract: We study the tradeoff between sample complexity and round complexity in on-demand sampling, where the learning algorithm adaptively samples from $k$ distributions over a limited number of rounds. In the realizable setting of Multi-Distribution Learning (MDL), we show that the optimal sample complexity of an $r$-round algorithm scales approximately as $dk^{\Theta(1/r)} / \epsilon$. For the general agnostic case, we present an algorithm that achieves near-optimal sample complexity of $\widetilde O((d + k) / \epsilon^2)$ within $\widetilde O(\sqrt{k})$ rounds. Of independent interest, we introduce a new framework, Optimization via On-Demand Sampling (OODS), which abstracts the sample-adaptivity tradeoff and captures most existing MDL algorithms. We establish nearly tight bounds on the round complexity in the OODS setting. The upper bounds directly yield the $\widetilde O(\sqrt{k})$-round algorithm for agnostic MDL, while the lower bounds imply that achieving sub-polynomial round complexity would require fundamentally new techniques that bypass the inherent hardness of OODS.
Authors: Yinan Yu, Samuel Scheidegger
Abstract: Runtime geofencing for ground vehicles is rapidly emerging as a critical technology for enforcing Operational Design Domains (ODDs). However, existing solutions struggle to reconcile high-fidelity learning with the structural requirements of verifiable control. We address this by introducing PCARNN-DCBF, a novel pipeline integrating a Physics-encoded Control-Affine Residual Neural Network with a preview-based Discrete Control Barrier Function. Unlike generic learned models, PCARNN explicitly preserves the control-affine structure of vehicle dynamics, ensuring the linearity required for reliable optimization. This enables the DCBF to enforce polygonal keep-in constraints via a real-time Quadratic Program (QP) that handles high relative degree and mitigates actuator saturation. Experiments in CARLA across electric and combustion platforms demonstrate that this structure-preserving approach significantly outperforms analytical and unstructured neural baselines.
Authors: Nils Wildt, Daniel M. Tartakovsky, Sergey Oladyshkin, Wolfgang Nowak
Abstract: Ordinary differential equations (ODEs) are a conventional way to describe the observed dynamics of physical systems. Scientists typically hypothesize about dynamical behavior, propose a mathematical model, and compare its predictions to data. However, modern computing and algorithmic advances now enable purely data-driven learning of governing dynamics directly from observations. In data-driven settings, one learns the ODE's right-hand side (RHS). Dense measurements are often assumed, yet high temporal resolution is typically both cumbersome and expensive. Consequently, one usually has only sparsely sampled data. In this work we introduce ChaosODE (CODE), a Polynomial Chaos ODE Expansion in which we use an arbitrary Polynomial Chaos Expansion (aPCE) for the ODE's right-hand side, resulting in a global orthonormal polynomial representation of dynamics. We evaluate the performance of CODE in several experiments on the Lotka-Volterra system, across varying noise levels, initial conditions, and predictions far into the future, even on previously unseen initial conditions. CODE exhibits remarkable extrapolation capabilities even when evaluated under novel initial conditions and shows advantages compared to well-examined methods using neural networks (NeuralODE) or kernel approximators (KernelODE) as the RHS representer. We observe that the high flexibility of NeuralODE and KernelODE degrades extrapolation capabilities under scarce data and measurement noise. Finally, we provide practical guidelines for robust optimization of dynamics-learning problems and illustrate them in the accompanying code.
Authors: Kim N. Nolle, Ivana Dusparic, Rhodri Cusack, Vinny Cahill
Abstract: Continual learning (CL) is a branch of machine learning that aims to enable agents to adapt and generalise previously learned abilities so that these can be reapplied to new tasks or environments. This is particularly useful in multi-task settings or in non-stationary environments, where the dynamics can change over time. This is particularly relevant in cyber-physical systems such as autonomous driving. However, despite recent advances in CL, successfully applying it to reinforcement learning (RL) is still an open problem. This paper highlights open challenges in continual RL (CRL) based on experiments in an autonomous driving environment. In this environment, the agent must learn to successfully park in four different scenarios corresponding to parking spaces oriented at varying angles. The agent is successively trained in these four scenarios one after another, representing a CL environment, using Proximal Policy Optimisation (PPO). These experiments exposed a number of open challenges in CRL: finding suitable abstractions of the environment, oversensitivity to hyperparameters, catastrophic forgetting, and efficient use of neural network capacity. Based on these identified challenges, we present open research questions that are important to be addressed for creating robust CRL systems. In addition, the identified challenges call into question the suitability of neural networks for CL. We also identify the need for interdisciplinary research, in particular between computer science and neuroscience.
Authors: Cheng Yin, Yankai Lin, Wang Xu, Sikyuen Tam, Xiangrui Zeng, Zhiyuan Liu, Zhouping Yin
Abstract: Enabling Vision-Language-Action (VLA) models to "think before acting" via Chain-of-Thought (CoT) is a promising path to overcoming the data-hungry nature of end-to-end robot policies. However, progress is stalled by a fundamental conflict: existing models use a single autoregressive decoder for both sequential CoT reasoning and high-dimensional, parallelizable robot actions. This architectural mismatch degrades motor control and fails to forge a strong causal link between thought and action. We introduce DeepThinkVLA, which resolves this conflict through a tightly integrated architecture and training strategy. Architecturally, our hybrid-attention decoder generates sequential CoT with causal attention and then switches to bidirectional attention for fast, parallel decoding of action vectors. This design is complemented by a two-stage training pipeline: we first use Supervised Fine-Tuning (SFT) to teach the model foundational reasoning, then apply Reinforcement Learning (RL) with task-success rewards to causally align the full reasoning-action sequence with desired outcomes. This synergy leads to state-of-the-art performance, achieving a 97.0% success rate on the LIBERO benchmark. Our ablations confirm the design's effectiveness: the hybrid architecture alone outperforms standard decoders by 15.5%, and the final RL stage provides a crucial 2% boost to secure top performance.
Authors: Michael McCabe, Payel Mukhopadhyay, Tanya Marwah, Bruno Regaldo-Saint Blancard, Francois Rozet, Cristiana Diaconu, Lucas Meyer, Kaze W. K. Wong, Hadi Sotoudeh, Alberto Bietti, Irina Espejo, Rio Fear, Siavash Golkar, Tom Hehir, Keiya Hirashima, Geraud Krawezik, Francois Lanusse, Rudy Morel, Ruben Ohana, Liam Parker, Mariel Pettee, Jeff Shen, Kyunghyun Cho, Miles Cranmer, Shirley Ho
Abstract: Foundation models have transformed machine learning for language and vision, but achieving comparable impact in physical simulation remains a challenge. Data heterogeneity and unstable long-term dynamics inhibit learning from sufficiently diverse dynamics, while varying resolutions and dimensionalities challenge efficient training on modern hardware. Through empirical and theoretical analysis, we incorporate new approaches to mitigate these obstacles, including a harmonic-analysis-based stabilization method, load-balanced distributed 2D and 3D training strategies, and compute-adaptive tokenization. Using these tools, we develop Walrus, a transformer-based foundation model developed primarily for fluid-like continuum dynamics. Walrus is pretrained on nineteen diverse scenarios spanning astrophysics, geoscience, rheology, plasma physics, acoustics, and classical fluids. Experiments show that Walrus outperforms prior foundation models on both short and long term prediction horizons on downstream tasks and across the breadth of pretraining data, while ablation studies confirm the value of our contributions to forecast stability, training throughput, and transfer performance over conventional approaches. Code and weights are released for community use.
Authors: Medha Kumar, Zifei Xu, Xin Wang, Tristan Webb
Abstract: Strong reasoning capabilities can now be achieved by large-scale reinforcement learning (RL) without any supervised fine-tuning. Although post-training quantization (PTQ) and quantization-aware training (QAT) are well studied in the context of fine-tuning, how quantization impacts RL in large reasoning models (LRMs) remains an open question. To answer this question, we conducted systematic experiments and discovered a significant gap in reasoning performance on mathematical benchmarks between post-RL quantized models and their quantization-aware RL optimized counterparts. Our findings suggest that quantization-aware RL training negatively impacted the learning process, whereas PTQ and QLoRA led to greater performance.
Authors: Yifan Xu, Vipul Gupta, Rohit Aggarwal, Varsha Mahadevan, Bhaskar Krishnamachari
Abstract: Retrieval-Augmented Generation (RAG) enhances large language models (LLMs) by pulling in external material, document, code, manuals, from vast and ever-growing corpora, to effectively answer user queries. The effectiveness of RAG depends significantly on aligning the number of retrieved documents with query characteristics: narrowly focused queries typically require fewer, highly relevant documents, whereas broader or ambiguous queries benefit from retrieving more extensive supporting information. However, the common static top-k retrieval approach fails to adapt to this variability, resulting in either insufficient context from too few documents or redundant information from too many. Motivated by these challenges, we introduce Cluster-based Adaptive Retrieval (CAR), an algorithm that dynamically determines the optimal number of documents by analyzing the clustering patterns of ordered query-document similarity distances. CAR detects the transition point within similarity distances, where tightly clustered, highly relevant documents shift toward less pertinent candidates, establishing an adaptive cut-off that scales with query complexity. On Coinbase's CDP corpus and the public MultiHop-RAG benchmark, CAR consistently picks the optimal retrieval depth and achieves the highest TES score, outperforming every fixed top-k baseline. In downstream RAG evaluations, CAR cuts LLM token usage by 60%, trims end-to-end latency by 22%, and reduces hallucinations by 10% while fully preserving answer relevance. Since integrating CAR into Coinbase's virtual assistant, we've seen user engagement jump by 200%.
Authors: S. K. Laha
Abstract: Forecasting nonlinear time series with multi-scale temporal structures remains a central challenge in complex systems modeling. We present a novel reservoir computing framework that combines delay embedding with random Fourier feature (RFF) mappings to capture such dynamics. Two formulations are investigated: a single-scale RFF reservoir, which employs a fixed kernel bandwidth, and a multi-scale RFF reservoir, which integrates multiple bandwidths to represent both fast and slow temporal dependencies. The framework is applied to a diverse set of canonical systems: neuronal models such as the Rulkov map, Izhikevich model, Hindmarsh-Rose model, and Morris-Lecar model, which exhibit spiking, bursting, and chaotic behaviors arising from fast-slow interactions; and ecological models including the predator-prey dynamics and Ricker map with seasonal forcing, which display multi-scale oscillations and intermittency. Across all cases, the multi-scale RFF reservoir consistently outperforms its single-scale counterpart, achieving lower normalized root mean square error (NRMSE) and more robust long-horizon predictions. These results highlight the effectiveness of explicitly incorporating multi-scale feature mappings into reservoir computing architectures for modeling complex dynamical systems with intrinsic fast-slow interactions.
Authors: Sourav De, Koustav Chowdhury, Bibhabasu Mandal, Sagar Ghosh, Swagatam Das, Debolina Paul, Saptarshi Chakraborty
Abstract: Clustering approaches that utilize convex loss functions have recently attracted growing interest in the formation of compact data clusters. Although classical methods like k-means and its wide family of variants are still widely used, all of them require the number of clusters k to be supplied as input, and many are notably sensitive to initialization. Convex clustering provides a more stable alternative by formulating the clustering task as a convex optimization problem, ensuring a unique global solution. However, it faces challenges in handling high-dimensional data, especially in the presence of noise and outliers. Additionally, strong fusion regularization, controlled by the tuning parameter, can hinder effective cluster formation within a convex clustering framework. To overcome these challenges, we introduce a robust approach that integrates convex clustering with the Median of Means (MoM) estimator, thus developing an outlier-resistant and efficient clustering framework that does not necessitate prior knowledge of the number of clusters. By leveraging the robustness of MoM alongside the stability of convex clustering, our method enhances both performance and efficiency, especially on large-scale datasets. Theoretical analysis demonstrates weak consistency under specific conditions, while experiments on synthetic and real-world datasets validate the method's superior performance compared to existing approaches.
Authors: Siyuan Li, Kai Yu, Anna Wang, Zicheng Liu, Chang Yu, Jingbo Zhou, Qirong Yang, Yucheng Guo, Xiaoming Zhang, Stan Z. Li
Abstract: Modeling genomic sequences faces two unsolved challenges: the information density varies widely across different regions, while there is no clearly defined minimum vocabulary unit. Relying on either four primitive bases or independently designed DNA tokenizers, existing approaches with naive masked language modeling pre-training often fail to adapt to the varying complexities of genomic sequences. Leveraging Token Merging techniques, this paper introduces a hierarchical architecture that jointly optimizes a dynamic genomic tokenizer and latent Transformers with context-aware pre-training tasks. As for network structures, the tokenization module automatically chunks adjacent bases into words by stacking multiple layers of the differentiable token merging blocks with local-window constraints, then a Latent Encoder captures the global context of these merged words by full-attention blocks. Symmetrically employing a Latent Decoder and a Local Decoder, MergeDNA learns with two pre-training tasks: Merged Token Reconstruction simultaneously trains the dynamic tokenization module and adaptively filters important tokens, while Adaptive Masked Token Modeling learns to predict these filtered tokens to capture informative contents. Extensive experiments show that MergeDNA achieves superior performance on three popular DNA benchmarks and several multi-omics tasks with fine-tuning or zero-shot evaluation, outperforming typical tokenization methods and large-scale DNA foundation models.
Authors: Jongyeon Yoon, Elyssa M. McMaster, Michael E. Kim, Gaurav Rudravaram, Kurt G. Schilling, Bennett A. Landman, Daniel Moyer
Abstract: Diffusion MRI (dMRI) provides a distinctive means to probe the microstructural architecture of living tissue, facilitating applications such as brain connectivity analysis, modeling across multiple conditions, and the estimation of macrostructural features. Tractography, which emerged in the final years of the 20th century and accelerated in the early 21st century, is a technique for visualizing white matter pathways in the brain using dMRI. Most diffusion tractography methods rely on procedural streamline propagators or global energy minimization methods. Although recent advancements in deep learning have enabled tasks that were previously challenging, existing tractography approaches are often non-differentiable, limiting their integration in end-to-end learning frameworks. While progress has been made in representing streamlines in differentiable frameworks, no existing method offers fully differentiable propagation. In this work, we propose a fully differentiable solution that retains numerical fidelity with a leading streamline algorithm. The key is that our PyTorch-engineered streamline propagator has no components that block gradient flow, making it fully differentiable. We show that our method matches standard propagators while remaining differentiable. By translating streamline propagation into a differentiable PyTorch framework, we enable deeper integration of tractography into deep learning workflows, laying the foundation for a new category of macrostructural reasoning that is not only computationally robust but also scientifically rigorous.
Authors: Peter Halmos, Boris Hanin
Abstract: Wasserstein gradient flow provides a general framework for minimizing an energy functional $J$ over the space of probability measures on a Riemannian manifold $(M,g)$. Its canonical time-discretization, the Jordan-Kinderlehrer-Otto (JKO) scheme, produces for any step size $\eta>0$ a sequence of probability distributions $\rho_k^\eta$ that approximate to first order in $\eta$ Wasserstein gradient flow on $J$. But the JKO scheme also has many other remarkable properties not shared by other first order integrators, e.g. it preserves energy dissipation and exhibits unconditional stability for $\lambda$-geodesically convex functionals $J$. To better understand the JKO scheme we characterize its implicit bias at second order in $\eta$. We show that $\rho_k^\eta$ are approximated to order $\eta^2$ by Wasserstein gradient flow on a \emph{modified} energy \[ J^{\eta}(\rho) = J(\rho) - \frac{\eta}{4}\int_M \Big\lVert \nabla_g \frac{\delta J}{\delta \rho} (\rho) \Big\rVert_{2}^{2} \,\rho(dx), \] obtained by subtracting from $J$ the squared metric curvature of $J$ times $\eta/4$. The JKO scheme therefore adds at second order in $\eta$ a \textit{deceleration} in directions where the metric curvature of $J$ is rapidly changing. This corresponds to canonical implicit biases for common functionals: for entropy the implicit bias is the Fisher information, for KL-divergence it is the Fisher-Hyv{\"a}rinen divergence, and for Riemannian gradient descent it is the kinetic energy in the metric $g$. To understand the differences between minimizing $J$ and $J^\eta$ we study \emph{JKO-Flow}, Wasserstein gradient flow on $J^\eta$, in several simple numerical examples. These include exactly solvable Langevin dynamics on the Bures-Wasserstein space and Langevin sampling from a quartic potential in 1D.
Authors: Marie Hein, Gregor Kasieczka, Michael Kr\"amer, Louis Moureaux, Alexander M\"uck, David Shih
Abstract: Anomaly detection has the potential to discover new physics in unexplored regions of the data. However, choosing the best anomaly detector for a given data set in a model-agnostic way is an important challenge which has hitherto largely been neglected. In this paper, we introduce the data-driven ARGOS metric, which has a sound theoretical foundation and is empirically shown to robustly select the most sensitive anomaly detection model given the data. Focusing on weakly-supervised, classifier-based anomaly detection methods, we show that the ARGOS metric outperforms other model selection metrics previously used in the literature, in particular the binary cross-entropy loss. We explore several realistic applications, including hyperparameter tuning as well as architecture and feature selection, and in all cases we demonstrate that ARGOS is robust to the noisy conditions of anomaly detection.
Authors: Xueying Ding, Xingyue Huang, Mingxuan Ju, Liam Collins, Yozen Liu, Leman Akoglu, Neil Shah, Tong Zhao
Abstract: Large language models produce powerful text embeddings, but their causal attention mechanism restricts the flow of information from later to earlier tokens, degrading representation quality. While recent methods attempt to solve this by prepending a single summary token, they over-compress information, hence harming performance on long documents. We propose Hierarchical Token Prepending (HTP), a method that resolves two critical bottlenecks. To mitigate attention-level compression, HTP partitions the input into blocks and prepends block-level summary tokens to subsequent blocks, creating multiple pathways for backward information flow. To address readout-level over-squashing, we replace last-token pooling with mean-pooling, a choice supported by theoretical analysis. HTP achieves consistent performance gains across 11 retrieval datasets and 30 general embedding benchmarks, especially in long-context settings. As a simple, architecture-agnostic method, HTP enhances both zero-shot and finetuned models, offering a scalable route to superior long-document embeddings.
Authors: Henry Wong, Clement Fung, Weiran Lin, Karen Li, Stanley Chen, Lujo Bauer
Abstract: To autonomously control vehicles, driving agents use outputs from a combination of machine-learning (ML) models, controller logic, and custom modules. Although numerous prior works have shown that adversarial examples can mislead ML models used in autonomous driving contexts, it remains unclear if these attacks are effective at producing harmful driving actions for various agents, environments, and scenarios. To assess the risk of adversarial examples to autonomous driving, we evaluate attacks against a variety of driving agents, rather than against ML models in isolation. To support this evaluation, we leverage CARLA, an urban driving simulator, to create and evaluate adversarial examples. We create adversarial patches designed to stop or steer driving agents, stream them into the CARLA simulator at runtime, and evaluate them against agents from the CARLA Leaderboard, a public repository of best-performing autonomous driving agents from an annual research competition. Unlike prior work, we evaluate attacks against autonomous driving systems without creating or modifying any driving-agent code and against all parts of the agent included with the ML model. We perform a case-study investigation of two attack strategies against three open-source driving agents from the CARLA Leaderboard across multiple driving scenarios, lighting conditions, and locations. Interestingly, we show that, although some attacks can successfully mislead ML models into predicting erroneous stopping or steering commands, some driving agents use modules, such as PID control or GPS-based rules, that can overrule attacker-manipulated predictions from ML models.
Authors: Michael T. Goodrich, Songyu Liu, Ioannis Panageas
Abstract: In this paper, we study the exact learning problem for weighted graphs, where we are given the vertex set, $V$, of a weighted graph, $G=(V,E,w)$, but we are not given $E$. The problem, which is also known as graph reconstruction, is to determine all the edges of $E$, including their weights, by asking queries about $G$ from an oracle. As we observe, using simple shortest-path length queries is not sufficient, in general, to learn a weighted graph. So we study a number of scenarios where it is possible to learn $G$ using a subquadratic number of composite queries, which combine two or three simple queries.
Authors: Pranav Indrakanti, Ivor Simpson
Abstract: We present an unsupervised single image bidirectional Magnetic Resonance Image (MRI) synthesizer that synthesizes an Ultra-Low Field (ULF) like image from a High-Field (HF) magnitude image and vice-versa. Unlike existing MRI synthesis models, our approach is inspired by the physics that drives contrast changes between HF and ULF MRIs. Our forward model simulates a HF to ULF transformation by estimating the tissue-type Signal-to-Noise ratio (SNR) values based on target contrast values. For the Super-Resolution task, we used an Implicit Neural Representation (INR) network to synthesize HF image by simultaneously predicting tissue-type segmentations and image intensity without observed HF data. The proposed method is evaluated using synthetic ULF-like data from generated from standard 3T T$_1$-weighted images for qualitative assessments and paired 3T-64mT T$_1$-weighted images for validation experiments. WM-GM contrast improved by 52% in synthetic ULF-like images and 37% in 64mT images. Sensitivity experiments demonstrated the robustness of our forward model to variations in target contrast, noise and initial seeding.
Authors: Geft\'e Almeida, Marcio Pohlmann, Alex Severo, Diego Kreutz, Tiago Heinrich, Louren\c{c}o Pereira
Abstract: In this study, we evaluate open-source models for security incident classification, comparing them with proprietary models. We utilize a dataset of anonymized real incidents, categorized according to the NIST SP 800-61r3 taxonomy and processed using five prompt-engineering techniques (PHP, SHP, HTP, PRP, and ZSL). The results indicate that, although proprietary models still exhibit higher accuracy, locally deployed open-source models provide advantages in privacy, cost-effectiveness, and data sovereignty.
Authors: Daniel Oliveira de Brito, Let\'icia Gabriella de Souza, Marcelo Matheus Gauy, Marcelo Finger, Arnaldo Candido Junior
Abstract: This technical report investigates the performance of pre-trained audio models on COVID-19 detection tasks using established benchmark datasets. We fine-tuned Audio-MAE and three PANN architectures (CNN6, CNN10, CNN14) on the Coswara and COUGHVID datasets, evaluating both intra-dataset and cross-dataset generalization. We implemented a strict demographic stratification by age and gender to prevent models from exploiting spurious correlations between demographic characteristics and COVID-19 status. Intra-dataset results showed moderate performance, with Audio-MAE achieving the strongest result on Coswara (0.82 AUC, 0.76 F1-score), while all models demonstrated limited performance on Coughvid (AUC 0.58-0.63). Cross-dataset evaluation revealed severe generalization failure across all models (AUC 0.43-0.68), with Audio-MAE showing strong performance degradation (F1-score 0.00-0.08). Our experiments demonstrate that demographic balancing, while reducing apparent model performance, provides more realistic assessment of COVID-19 detection capabilities by eliminating demographic leakage - a confounding factor that inflate performance metrics. Additionally, the limited dataset sizes after balancing (1,219-2,160 samples) proved insufficient for deep learning models that typically require substantially larger training sets. These findings highlight fundamental challenges in developing generalizable audio-based COVID-19 detection systems and underscore the importance of rigorous demographic controls for clinically robust model evaluation.
Authors: Mohamed Abdallah Salem, Hamdy Ahmed Ashour, Ahmed Elshenawy
Abstract: This research addresses the significant challenges of energy consumption and environmental impact in laser cutting by proposing novel deep learning (DL) methodologies to achieve energy reduction. Recognizing the current lack of adaptive control and the open-loop nature of CO2 laser suction pumps, this study utilizes closed-loop configurations that dynamically adjust pump power based on both the material being cut and the smoke level generated. To implement this adaptive system, diverse material classification methods are introduced, including techniques leveraging lens-less speckle sensing with a customized Convolutional Neural Network (CNN) and an approach using a USB camera with transfer learning via the pre-trained VGG16 CNN model. Furthermore, a separate DL model for smoke level detection is employed to simultaneously refine the pump's power output. This integration prompts the exhaust suction pump to automatically halt during inactive times and dynamically adjust power during operation, leading to experimentally proven and remarkable energy savings, with results showing a 20% to 50% reduction in the smoke suction pump's energy consumption, thereby contributing substantially to sustainable development in the manufacturing sector.
Authors: Joey Velez-Ginorio, Nada Amin, Konrad Kording, Steve Zdancewic
Abstract: Discrete structures are currently second-class in differentiable programming. Since functions over discrete structures lack overt derivatives, differentiable programs do not differentiate through them and limit where they can be used. For example, when programming a neural network, conditionals and iteration cannot be used everywhere; they can break the derivatives necessary for gradient-based learning to work. This limits the class of differentiable algorithms we can directly express, imposing restraints on how we build neural networks and differentiable programs more generally. However, these restraints are not fundamental. Recent work shows conditionals can be first-class, by compiling them into differentiable form as linear neurons. Similarly, this work shows iteration can be first-class -- by compiling to linear recurrent neurons. We present a minimal typed, higher-order and linear programming language with iteration called $\textsf{Cajal}\scriptstyle(\mathbb{\multimap}, \mathbb{2}, \mathbb{N})$. We prove its programs compile correctly to recurrent neurons, allowing discrete algorithms to be expressed in a differentiable form compatible with gradient-based learning. With our implementation, we conduct two experiments where we link these recurrent neurons against a neural network solving an iterative image transformation task. This determines part of its function prior to learning. As a result, the network learns faster and with greater data-efficiency relative to a neural network programmed without first-class iteration. A key lesson is that recurrent neurons enable a rich interplay between learning and the discrete structures of ordinary programming.
Authors: Haizhou Wen, He Li, Zhen Li
Abstract: Reconstruction of 3D erythrocyte or red blood cell (RBC) morphology from partial observations, such as microscope images, is essential for understanding the physiology of RBC aging and the pathology of various RBC disorders. In this study, we propose a multi-fidelity neural network (MFNN) approach to fuse high-fidelity cross-sections of an RBC, with a morphologically similar low-fidelity reference 3D RBC shape to recover its full 3D surface. The MFNN predictor combines a convolutional neural network trained on low-fidelity reference RBC data with a feedforward neural network that captures nonlinear morphological correlations, and augments training with surface area and volume constraints for regularization in the low-fidelity branch. This approach is theoretically grounded by a topological homeomorphism between a sphere and 3D RBC surfaces, with training data generated by dissipative particle dynamics simulations of stomatocyte-discocyte-echinocyte transformation. Benchmarking across diverse RBC shapes observed in normal and aged populations, our results show that the MFNN predictor can reconstruct complex RBC morphologies with over 95% coordinate accuracy when provided with at least two orthogonal cross-sections. It is observed that informative oblique cross-sections intersecting spicule tips of echinocytes improve both local and global feature reconstruction, highlighting the value of feature-aware sampling. Our study further evaluates the influence of sampling strategies, shape dissimilarity, and noise, showing enhanced robustness under physically constrained training. Altogether, these results demonstrate the capability of MFNN to reconstruct the 3D shape of normal and aged RBCs from partial cross-sections as observed in conventional microscope images, which could facilitate the quantitative analysis of RBC morphological parameters in normal and disease-related RBC samples.
Authors: Basel Shbita, Farhan Ahmed, Chad DeLuca
Abstract: Large language models (LLMs) have demonstrated excellent capabilities in generating structured diagrams from natural language descriptions. In particular, they have shown great promise in generating sequence diagrams for software engineering, typically represented in a text-based syntax such as Mermaid. However, systematic evaluations in this space remain underdeveloped as there is a lack of existing benchmarks to assess the LLM's correctness in this task. To address this shortcoming, we introduce MermaidSeqBench, a human-verified and LLM-synthetically-extended benchmark for assessing an LLM's capabilities in generating Mermaid sequence diagrams from textual prompts. The benchmark consists of a core set of 132 samples, starting from a small set of manually crafted and verified flows. These were expanded via a hybrid methodology combining human annotation, in-context LLM prompting, and rule-based variation generation. Our benchmark uses an LLM-as-a-judge model to assess Mermaid sequence diagram generation across fine-grained metrics, including syntax correctness, activation handling, error handling, and practical usability. We perform initial evaluations on numerous state-of-the-art LLMs and utilize multiple LLM judge models to demonstrate the effectiveness and flexibility of our benchmark. Our results reveal significant capability gaps across models and evaluation modes. Our proposed benchmark provides a foundation for advancing research in structured diagram generation and for developing more rigorous, fine-grained evaluation methodologies.
Authors: Zanxu Wang, Homayoon Beigi
Abstract: This paper addresses data quality issues in multimodal emotion recognition in conversation (MERC) through systematic quality control and multi-stage transfer learning. We implement a quality control pipeline for MELD and IEMOCAP datasets that validates speaker identity, audio-text alignment, and face detection. We leverage transfer learning from speaker and face recognition, assuming that identity-discriminative embeddings capture not only stable acoustic and Facial traits but also person-specific patterns of emotional expression. We employ RecoMadeEasy(R) engines for extracting 512-dimensional speaker and face embeddings, fine-tune MPNet-v2 for emotion-aware text representations, and adapt these features through emotion-specific MLPs trained on unimodal datasets. MAMBA-based trimodal fusion achieves 64.8% accuracy on MELD and 74.3% on IEMOCAP. These results show that combining identity-based audio and visual embeddings with emotion-tuned text representations on a quality-controlled subset of data yields consistent competitive performance for multimodal emotion recognition in conversation and provides a basis for further improvement on challenging, low-frequency emotion classes.
Authors: Ahmet Umur \"Ozsoy
Abstract: Calibration of option pricing models is routinely repeated as markets evolve, yet modern systems lack an operator for removing data from a calibrated model without full retraining. When quotes become stale, corrupted, or subject to deletion requirements, existing calibration pipelines must rebuild the entire nonlinear least-squares problem, even if only a small subset of data must be excluded. In this work, we introduce a principled framework for selective forgetting (machine unlearning) in parametric option calibration. We provide stability guarantees, perturbation bounds, and show that the proposed operators satisfy local exactness under standard regularity assumptions.
Authors: Nicholas Cooper, Lijun Chen, Sailesh Dwivedy, Danna Gurari
Abstract: Knowledge distillation (KD) methods can transfer knowledge of a parameter-heavy teacher model to a light-weight student model. The status quo for feature KD methods is to utilize loss functions based on logits (i.e., pre-softmax class scores) and intermediate layer features (i.e., latent representations). Unlike previous approaches, we propose a feature KD framework for training the student's backbone using feature-based losses exclusively (i.e., without logit-based losses such as cross entropy). Leveraging recent discoveries about the geometry of latent representations, we introduce a knowledge quality metric for identifying which teacher layers provide the most effective knowledge for distillation. Experiments on three image classification datasets with four diverse student-teacher pairs, spanning convolutional neural networks and vision transformers, demonstrate our KD method achieves state-of-the-art performance, delivering top-1 accuracy boosts of up to 15% over standard approaches. We publically share our code to facilitate future work at https://github.com/Thegolfingocto/KD_wo_CE.
Authors: Vladimir Arkhipkin, Vladimir Korviakov, Nikolai Gerasimenko, Denis Parkhomenko, Viacheslav Vasilev, Alexey Letunovskiy, Maria Kovaleva, Nikolai Vaulin, Ivan Kirillov, Lev Novitskiy, Denis Koposov, Nikita Kiselev, Alexander Varlamov, Dmitrii Mikhailov, Vladimir Polovnikov, Andrey Shutkin, Ilya Vasiliev, Julia Agafonova, Anastasiia Kargapoltseva, Anna Dmitrienko, Anastasia Maltseva, Anna Averchenkova, Olga Kim, Tatiana Nikulina, Denis Dimitrov
Abstract: This report introduces Kandinsky 5.0, a family of state-of-the-art foundation models for high-resolution image and 10-second video synthesis. The framework comprises three core line-up of models: Kandinsky 5.0 Image Lite - a line-up of 6B parameter image generation models, Kandinsky 5.0 Video Lite - a fast and lightweight 2B parameter text-to-video and image-to-video models, and Kandinsky 5.0 Video Pro - 19B parameter models that achieves superior video generation quality. We provide a comprehensive review of the data curation lifecycle - including collection, processing, filtering and clustering - for the multi-stage training pipeline that involves extensive pre-training and incorporates quality-enhancement techniques such as self-supervised fine-tuning (SFT) and reinforcement learning (RL)-based post-training. We also present novel architectural, training, and inference optimizations that enable Kandinsky 5.0 to achieve high generation speeds and state-of-the-art performance across various tasks, as demonstrated by human evaluation. As a large-scale, publicly available generative framework, Kandinsky 5.0 leverages the full potential of its pre-training and subsequent stages to be adapted for a wide range of generative applications. We hope that this report, together with the release of our open-source code and training checkpoints, will substantially advance the development and accessibility of high-quality generative models for the research community.
Authors: Fatemeh Lotfi, Hossein Rajoli, Fatemeh Afghah
Abstract: Next-generation networks utilize the Open Radio Access Network (O-RAN) architecture to enable dynamic resource management, facilitated by the RAN Intelligent Controller (RIC). While deep reinforcement learning (DRL) models show promise in optimizing network resources, they often struggle with robustness and generalizability in dynamic environments. This paper introduces a novel resource management approach that enhances the Soft Actor Critic (SAC) algorithm with Sharpness-Aware Minimization (SAM) in a distributed Multi-Agent RL (MARL) framework. Our method introduces an adaptive and selective SAM mechanism, where regularization is explicitly driven by temporal-difference (TD)-error variance, ensuring that only agents facing high environmental complexity are regularized. This targeted strategy reduces unnecessary overhead, improves training stability, and enhances generalization without sacrificing learning efficiency. We further incorporate a dynamic $\rho$ scheduling scheme to refine the exploration-exploitation trade-off across agents. Experimental results show our method significantly outperforms conventional DRL approaches, yielding up to a $22\%$ improvement in resource allocation efficiency and ensuring superior QoS satisfaction across diverse O-RAN slices.
Authors: Reza Mirjalili, Behrad Braghi, Shahram Shadrokh Sikari
Abstract: Accurate prediction of project duration and cost remains one of the most challenging aspects of project management, particularly in resource-constrained and interdependent task networks. Traditional analytical techniques such as the Critical Path Method (CPM) and Program Evaluation and Review Technique (PERT) rely on simplified and often static assumptions regarding task interdependencies and resource performance. This study proposes a novel resource-based predictive framework that integrates network representations of project activities with graph neural networks (GNNs) to capture structural and contextual relationships among tasks, resources, and time-cost dynamics. The model represents the project as a heterogeneous activity-resource graph in which nodes denote activities and resources, and edges encode temporal and resource dependencies. We evaluate multiple learning paradigms, including GraphSAGE and Temporal Graph Networks, on both synthetic and benchmark project datasets. Experimental results show that the proposed GNN framework achieves an average 23 to 31 percent reduction in mean absolute error compared to traditional regression and tree-based methods, while improving the coefficient of determination R2 from approximately 0.78 to 0.91 for large and complex project networks. Furthermore, the learned embeddings provide interpretable insights into resource bottlenecks and critical dependencies, enabling more explainable and adaptive scheduling decisions.
Authors: Katie Rainey, Erin Hausmann, Donald Waagen, David Gray, Donald Hulsey
Abstract: Understanding the relationships between data points in the latent decision space derived by the deep learning system is critical to evaluating and interpreting the performance of the system on real world data. Detecting \textit{out-of-distribution} (OOD) data for deep learning systems continues to be an active research topic. We investigate the connection between latent space OOD detection and classification accuracy of the model. Using open source simulated and measured Synthetic Aperture RADAR (SAR) datasets, we empirically demonstrate that the OOD detection cannot be used as a proxy measure for model performance. We hope to inspire additional research into the geometric properties of the latent space that may yield future insights into deep learning robustness and generalizability.
Authors: Kexin Chu, Dawei Xiang, Zixu Shen, Yiwei Yang, Zecheng Liu, Wei Zhang
Abstract: Mixture-of-Experts (MoE) models scale LLM capacity efficiently, but deployment on consumer GPUs is limited by the large memory footprint of inactive experts. Static post-training quantization reduces storage costs but cannot adapt to shifting activation patterns, causing accuracy loss under aggressive compression. So we present DynaExq, a runtime system that treats expert precision as a first-class, dynamically managed resource. DynaExq combines (1) a hotness-aware precision controller that continuously aligns expert bit-widths with long-term activation statistics, (2) a fully asynchronous precision-switching pipeline that overlaps promotion and demotion with MoE computation, and (3) a fragmentation-free memory pooling mechanism that supports hybrid-precision experts with deterministic allocation. Together, these components enable stable, non-blocking precision transitions under strict HBM budgets. Across Qwen3-30B and Qwen3-80B MoE models and six representative benchmarks, DynaExq deploys large LLMs on single RTX 5090 and A6000 GPUs and improves accuracy by up to 4.03 points over static low-precision baselines. The results show that adaptive, workload-aware quantization is an effective strategy for memory-constrained MoE serving.
Authors: Yicheng Zhan, Xiangjun Gao, Long Quan, Kaan Ak\c{s}it
Abstract: We propose a new hologram representation based on structured complex-valued 2D Gaussian primitives, which replaces per-pixel information storage and reduces the parameter search space by up to 10:1. To enable end-to-end training, we develop a differentiable rasterizer for our representation, integrated with a GPU-optimized light propagation kernel in free space. Our extensive experiments show that our method achieves up to 2.5x lower VRAM usage and 50% faster optimization while producing higher-fidelity reconstructions than existing methods. We further introduce a conversion procedure that adapts our representation to practical hologram formats, including smooth and random phase-only holograms. Our experiments show that this procedure can effectively suppress noise artifacts observed in previous methods. By reducing the hologram parameter search space, our representation enables a more scalable hologram estimation in the next-generation computer-generated holography systems.
Authors: Jian-Ting Guo, Yu-Cheng Chen, Ping-Chun Hsieh, Kuo-Hao Ho, Po-Wei Huang, Ti-Rong Wu, I-Chen Wu
Abstract: Human-like agents have long been one of the goals in pursuing artificial intelligence. Although reinforcement learning (RL) has achieved superhuman performance in many domains, relatively little attention has been focused on designing human-like RL agents. As a result, many reward-driven RL agents often exhibit unnatural behaviors compared to humans, raising concerns for both interpretability and trustworthiness. To achieve human-like behavior in RL, this paper first formulates human-likeness as trajectory optimization, where the objective is to find an action sequence that closely aligns with human behavior while also maximizing rewards, and adapts the classic receding-horizon control to human-like learning as a tractable and efficient implementation. To achieve this, we introduce Macro Action Quantization (MAQ), a human-like RL framework that distills human demonstrations into macro actions via Vector-Quantized VAE. Experiments on D4RL Adroit benchmarks show that MAQ significantly improves human-likeness, increasing trajectory similarity scores, and achieving the highest human-likeness rankings among all RL agents in the human evaluation study. Our results also demonstrate that MAQ can be easily integrated into various off-the-shelf RL algorithms, opening a promising direction for learning human-like RL agents. Our code is available at https://rlg.iis.sinica.edu.tw/papers/MAQ.
Authors: Haodong Chen, Guido Zuccon, Teerapong Leelanupab
Abstract: Genomic question answering often requires complex reasoning and integration across diverse biomedical sources. GeneGPT addressed this challenge by combining domain-specific APIs with OpenAI's code-davinci-002 large language model to enable natural language interaction with genomic databases. However, its reliance on a proprietary model limits scalability, increases operational costs, and raises concerns about data privacy and generalization. In this work, we revisit and reproduce GeneGPT in a pilot study using open source models, including Llama 3.1, Qwen2.5, and Qwen2.5 Coder, within a monolithic architecture; this allows us to identify the limitations of this approach. Building on this foundation, we then develop OpenBioLLM, a modular multi-agent framework that extends GeneGPT by introducing agent specialization for tool routing, query generation, and response validation. This enables coordinated reasoning and role-based task execution. OpenBioLLM matches or outperforms GeneGPT on over 90% of the benchmark tasks, achieving average scores of 0.849 on Gene-Turing and 0.830 on GeneHop, while using smaller open-source models without additional fine-tuning or tool-specific pretraining. OpenBioLLM's modular multi-agent design reduces latency by 40-50% across benchmark tasks, significantly improving efficiency without compromising model capability. The results of our comprehensive evaluation highlight the potential of open-source multi-agent systems for genomic question answering. Code and resources are available at https://github.com/ielab/OpenBioLLM.
Authors: Khaled Hamidouche (NVIDIA Corporation), John Bachan (NVIDIA Corporation), Pak Markthub (NVIDIA Corporation), Peter-Jan Gootzen (NVIDIA Corporation), Elena Agostini (NVIDIA Corporation), Sylvain Jeaugey (NVIDIA Corporation), Aamir Shafi (NVIDIA Corporation), Georgios Theodorakis (NVIDIA Corporation), Manjunath Gorentla Venkata (NVIDIA Corporation)
Abstract: Modern AI workloads, especially Mixture-of-Experts (MoE) architectures, increasingly demand low-latency, fine-grained GPU-to-GPU communication with device-side control. Traditional GPU communication follows a host-initiated model, where the CPU orchestrates all communication operations - a characteristic of the CUDA runtime. Although robust for collective operations, applications requiring tight integration of computation and communication can benefit from device-initiated communication that eliminates CPU coordination overhead. NCCL 2.28 introduces the Device API with three operation modes: Load/Store Accessible (LSA) for NVLink/PCIe, Multimem for NVLink SHARP, and GPU-Initiated Networking (GIN) for network RDMA. This paper presents the GIN architecture, design, semantics, and highlights its impact on MoE communication. GIN builds on a three-layer architecture: i) NCCL Core host-side APIs for device communicator setup and collective memory window registration; ii) Device-side APIs for remote memory operations callable from CUDA kernels; and iii) A network plugin architecture with dual semantics (GPUDirect Async Kernel-Initiated and Proxy) for broad hardware support. The GPUDirect Async Kernel-Initiated backend leverages DOCA GPUNetIO for direct GPU-to-NIC communication, while the Proxy backend provides equivalent functionality via lock-free GPU-to-CPU queues over standard RDMA networks. We demonstrate GIN's practicality through integration with DeepEP, an MoE communication library. Comprehensive benchmarking shows that GIN provides device-initiated communication within NCCL's unified runtime, combining low-latency operations with NCCL's collective algorithms and production infrastructure.
Authors: Bohan Zhang, Zihao Wang, Hengyu Fu, Jason D. Lee
Abstract: In deep learning, a central issue is to understand how neural networks efficiently learn high-dimensional features. To this end, we explore the gradient descent learning of a general Gaussian Multi-index model $f(\boldsymbol{x})=g(\boldsymbol{U}\boldsymbol{x})$ with hidden subspace $\boldsymbol{U}\in \mathbb{R}^{r\times d}$, which is the canonical setup to study representation learning. We prove that under generic non-degenerate assumptions on the link function, a standard two-layer neural network trained via layer-wise gradient descent can agnostically learn the target with $o_d(1)$ test error using $\widetilde{\mathcal{O}}(d)$ samples and $\widetilde{\mathcal{O}}(d^2)$ time. The sample and time complexity both align with the information-theoretic limit up to leading order and are therefore optimal. During the first stage of gradient descent learning, the proof proceeds via showing that the inner weights can perform a power-iteration process. This process implicitly mimics a spectral start for the whole span of the hidden subspace and eventually eliminates finite-sample noise and recovers this span. It surprisingly indicates that optimal results can only be achieved if the first layer is trained for more than $\mathcal{O}(1)$ steps. This work demonstrates the ability of neural networks to effectively learn hierarchical functions with respect to both sample and time efficiency.
Authors: Nishchala Thakur, Swati Kochhar, Deepti R. Bathula, Sukrit Gupta
Abstract: Active learning reduces annotation costs in medical imaging by strategically selecting the most informative samples for labeling. However, individual acquisition strategies often exhibit inconsistent behavior across different stages of the active learning cycle. We propose Cyclical and Performance-Adaptive Multi-Strategy Active Learning (WaveFuse-AL), a novel framework that adaptively fuses multiple established acquisition strategies-BALD, BADGE, Entropy, and CoreSet throughout the learning process. WaveFuse-AL integrates cyclical (sinusoidal) temporal priors with performance-driven adaptation to dynamically adjust strategy importance over time. We evaluate WaveFuse-AL on three medical imaging benchmarks: APTOS-2019 (multi-class classification), RSNA Pneumonia Detection (binary classification), and ISIC-2018 (skin lesion segmentation). Experimental results demonstrate that WaveFuse-AL consistently outperforms both single-strategy and alternating-strategy baselines, achieving statistically significant performance improvements (on ten out of twelve metric measurements) while maximizing the utility of limited annotation budgets.
Authors: Amit Kumar, Maninder Kaur, Raghvendra Mall, Sukrit Gupta
Abstract: Spatial Transcriptomics enables mapping of gene expression within its native tissue context, but current platforms measure only a limited set of genes due to experimental constraints and excessive costs. To overcome this, computational models integrate Single-Cell RNA Sequencing data with Spatial Transcriptomics to predict unmeasured genes. We propose CASPER, a cross-attention based framework that predicts unmeasured gene expression in Spatial Transcriptomics by leveraging centroid-level representations from Single-Cell RNA Sequencing. We performed rigorous testing over four state-of-the-art Spatial Transcriptomics/Single-Cell RNA Sequencing dataset pairs across four existing baseline models. CASPER shows significant improvement in nine out of the twelve metrics for our experiments. This work paves the way for further work in Spatial Transcriptomics to Single-Cell RNA Sequencing modality translation. The code for CASPER is available at https://github.com/AI4Med-Lab/CASPER.
Authors: Eugene Ndiaye
Abstract: Conformal prediction (CP) constructs uncertainty sets for model outputs with finite-sample coverage guarantees. A candidate output is included in the prediction set if its non-conformity score is not considered extreme relative to the scores observed on a set of calibration examples. However, this procedure is only straightforward when scores are scalar-valued, which has limited CP to real-valued scores or ad-hoc reductions to one dimension. The problem of ordering vectors has been studied via optimal transport (OT), which provides a principled method for defining vector-ranks and multivariate quantile regions, though typically with only asymptotic coverage guarantees. We restore finite-sample, distribution-free coverage by conformalizing the vector-valued OT quantile region. Here, a candidate's rank is defined via a transport map computed for the calibration scores augmented with that candidate's score. This defines a continuum of OT problems for which we prove that the resulting optimal assignment is piecewise-constant across a fixed polyhedral partition of the score space. This allows us to characterize the entire prediction set tractably, and provides the machinery to address a deeper limitation of prediction sets: that they only indicate which outcomes are plausible, but not their relative likelihood. In one dimension, conformal predictive distributions (CPDs) fill this gap by producing a predictive distribution with finite-sample calibration. Extending CPDs beyond one dimension remained an open problem. We construct, to our knowledge, the first multivariate CPDs with finite-sample calibration, i.e., they define a valid multivariate distribution where any derived uncertainty region automatically has guaranteed coverage. We present both conservative and exact randomized versions, the latter resulting in a multivariate generalization of the classical Dempster-Hill procedure.
Authors: Meihua Zhou, Xinyu Tong, Jiarui Zhao, Min Cheng, Li Yang, Lei Tian, Nan Wan
Abstract: High-dimensional neuroimaging analyses for clinical diagnosis are often constrained by compromises in spatiotemporal fidelity and by the limited adaptability of large-scale, general-purpose models. To address these challenges, we introduce Dynamic Curriculum Learning for Spatiotemporal Encoding (DCL-SE), an end-to-end framework centered on data-driven spatiotemporal encoding (DaSE). We leverage Approximate Rank Pooling (ARP) to efficiently encode three-dimensional volumetric brain data into information-rich, two-dimensional dynamic representations, and then employ a dynamic curriculum learning strategy, guided by a Dynamic Group Mechanism (DGM), to progressively train the decoder, refining feature extraction from global anatomical structures to fine pathological details. Evaluated across six publicly available datasets, including Alzheimer's disease and brain tumor classification, cerebral artery segmentation, and brain age prediction, DCL-SE consistently outperforms existing methods in accuracy, robustness, and interpretability. These findings underscore the critical importance of compact, task-specific architectures in the era of large-scale pretrained networks.
Authors: Firdavs Nasriddinov, Rafal Kocielnik, Anima Anandkumar, Andrew J. Hung
Abstract: High-quality intraoperative feedback from a surgical trainer is pivotal for improving trainee performance and long-term skill acquisition. Automating natural, trainer-style feedback promises timely, accessible, and consistent guidance at scale but requires models that understand clinically relevant representations. We present a structure-aware pipeline that learns a surgical action ontology from real trainer-to-trainee transcripts (33 surgeries) and uses it to condition feedback generation. We contribute by (1) mining Instrument-Action-Target (IAT) triplets from real-world feedback text and clustering surface forms into normalized categories, (2) fine-tuning a video-to-IAT model that leverages the surgical procedure and task contexts as well as fine-grained temporal instrument motion, and (3) demonstrating how to effectively use IAT triplet representations to guide GPT-4o in generating clinically grounded, trainer-style feedback. We show that, on Task 1: Video-to-IAT recognition, our context injection and temporal tracking deliver consistent AUC gains (Instrument: 0.67 to 0.74; Action: 0.60 to 0.63; Tissue: 0.74 to 0.79). For Task 2: feedback text generation (rated on a 1-5 fidelity rubric where 1 = opposite/unsafe, 3 = admissible, and 5 = perfect match to a human trainer), GPT-4o from video alone scores 2.17, while IAT conditioning reaches 2.44 (+12.4%), doubling the share of admissible generations with score >= 3 from 21% to 42%. Traditional text-similarity metrics also improve: word error rate decreases by 15-31% and ROUGE (phrase/substring overlap) increases by 9-64%. Grounding generation in explicit IAT structure improves fidelity and yields clinician-verifiable rationales, supporting auditable use in surgical training.
Authors: Ahmed Aboulfotouh, Hatem Abou-Zeid
Abstract: Wireless foundation models (WFMs) have recently demonstrated promising capabilities, jointly performing multiple wireless functions and adapting effectively to new environments. However, while current WFMs process only one modality, depending on the task and operating conditions, the most informative modality changes and no single modality is best for all tasks. WFMs should therefore be designed to accept multiple modalities to enable a broader and more diverse range of tasks and scenarios. In this work, we propose and build the first multimodal wireless foundation model capable of processing both raw IQ streams and image-like wireless modalities (e.g., spectrograms and CSI) and performing multiple tasks across both. We introduce masked wireless modeling for the multimodal setting, a self-supervised objective and pretraining recipe that learns a joint representation from IQ streams and image-like wireless modalities. We evaluate the model on five tasks across both modality families: image-based (human activity sensing, RF signal classification, 5G NR positioning) and IQ-based (RF device fingerprinting, interference detection/classification). The multimodal WFM is competitive with single-modality WFMs, and in several cases surpasses their performance. Our results demonstrates the strong potential of developing multimodal WFMs that support diverse wireless tasks across different modalities. We believe this provides a concrete step toward both AI-native 6G and the vision of joint sensing, communication, and localization.
Authors: Yang Wu, Rujing Yao, Tong Zhang, Yufei Shi, Zhuoren Jiang, Zhushan Li, Xiaozhong Liu
Abstract: Large Language Models (LLMs) are increasingly integrated into intelligent tutoring systems to provide human-like and adaptive instruction. However, most existing approaches fail to capture how students' knowledge evolves dynamically across their proficiencies, conceptual gaps, and forgetting patterns. This challenge is particularly acute in mathematics tutoring, where effective instruction requires fine-grained scaffolding precisely calibrated to each student's mastery level and cognitive retention. To address this issue, we propose TASA (Teaching According to Students' Aptitude), a student-aware tutoring framework that integrates persona, memory, and forgetting dynamics for personalized mathematics learning. Specifically, TASA maintains a structured student persona capturing proficiency profiles and an event memory recording prior learning interactions. By incorporating a continuous forgetting curve with knowledge tracing, TASA dynamically updates each student's mastery state and generates contextually appropriate, difficulty-calibrated questions and explanations. Empirical results demonstrate that TASA achieves superior learning outcomes and more adaptive tutoring behavior compared to representative baselines, underscoring the importance of modeling temporal forgetting and learner profiles in LLM-based tutoring systems.
Authors: Jun Hyeun Kang, Jung Eek Son, Tae In Ahn
Abstract: The application of spectral-shifting films in greenhouses to shift green light to red light has shown variable growth responses across crop species. However, the yield enhancement of crops under altered light quality is related to the collective effects of the specific biophysical characteristics of each species. Considering only one attribute of a crop has limitations in understanding the relationship between sunlight quality adjustments and crop growth performance. Therefore, this study aims to comprehensively link multiple plant phenotypic traits and daily light integral considering the physiological responses of crops to their growth outcomes under SF using artificial intelligence. Between 2021 and 2024, various leafy, fruiting, and root crops were grown in greenhouses covered with either PEF or SF, and leaf reflectance, leaf mass per area, chlorophyll content, daily light integral, and light saturation point were measured from the plants cultivated in each condition. 210 data points were collected, but there was insufficient data to train deep learning models, so a variational autoencoder was used for data augmentation. Most crop yields showed an average increase of 22.5% under SF. These data were used to train several models, including logistic regression, decision tree, random forest, XGBoost, and feedforward neural network (FFNN), aiming to binary classify whether there was a significant effect on yield with SF application. The FFNN achieved a high classification accuracy of 91.4% on a test dataset that was not used for training. This study provide insight into the complex interactions between leaf phenotypic and photosynthetic traits, environmental conditions, and solar spectral components by improving the ability to predict solar spectral shift effects using SF.
Authors: Rishikant Chigrupaatii, Ponnada Sai Tulasi Kanishka, Lalit Chandra Routhu, Martin Patel Sama Supratheek Reddy, Divyam Gupta, Dasari Srikar, Krishna Teja Kuchimanchi, Rajiv Misra, Rohun Tripathi
Abstract: With nearly 1.5 billion people and more than 120 major languages, India represents one of the most diverse regions in the world. As multilingual Vision-Language Models (VLMs) gain prominence, robust evaluation methodologies are essential to drive progress toward equitable AI for low-resource languages. Current multilingual VLM evaluations suffer from four major limitations: reliance on unverified auto-translations, narrow task/domain coverage, limited sample sizes, and lack of cultural and natively sourced Question-Answering (QA). To address these gaps, we present a scalable framework to evaluate VLMs in Indian languages and compare it with performance in English. Using the framework, we generate HinTel-AlignBench, a benchmark that draws from diverse sources in Hindi and Telugu with English-aligned samples. Our contributions are threefold: (1) a semi-automated dataset creation framework combining back-translation, filtering, and human verification; (2) the most comprehensive vision-language benchmark for Hindi and and Telugu, including adapted English datasets (VQAv2, RealWorldQA, CLEVR-Math) and native novel Indic datasets (JEE for STEM, VAANI for cultural grounding) with approximately 4,000 QA pairs per language; and (3) a detailed performance analysis of various State-of-the-Art (SOTA) open-weight and closed-source VLMs. We find a regression in performance for tasks in English versus in Indian languages for 4 out of 5 tasks across all the models, with an average regression of 8.3 points in Hindi and 5.5 points for Telugu. We categorize common failure modes to highlight concrete areas of improvement in multilingual multimodal understanding.
Authors: Wasif Jalal, Md Nafiu Rahman, M. Sohel Rahman
Abstract: Accurate brain age estimation from structural MRI is a valuable biomarker for studying aging and neurodegeneration. Traditional regression and CNN-based methods face limitations such as manual feature engineering, limited receptive fields, and overfitting on heterogeneous data. Pure transformer models, while effective, require large datasets and high computational cost. We propose Brain ResNet over trained Vision Transformer (BrainRotViT), a hybrid architecture that combines the global context modeling of vision transformers (ViT) with the local refinement of residual CNNs. A ViT encoder is first trained on an auxiliary age and sex classification task to learn slice-level features. The frozen encoder is then applied to all sagittal slices to generate a 2D matrix of embedding vectors, which is fed into a residual CNN regressor that incorporates subject sex at the final fully-connected layer to estimate continuous brain age. Our method achieves an MAE of 3.34 years (Pearson $r=0.98$, Spearman $\rho=0.97$, $R^2=0.95$) on validation across 11 MRI datasets encompassing more than 130 acquisition sites, outperforming baseline and state-of-the-art models. It also generalizes well across 4 independent cohorts with MAEs between 3.77 and 5.04 years. Analyses on the brain age gap (the difference between the predicted age and actual age) show that aging patterns are associated with Alzheimer's disease, cognitive impairment, and autism spectrum disorder. Model attention maps highlight aging-associated regions of the brain, notably the cerebellar vermis, precentral and postcentral gyri, temporal lobes, and medial superior frontal gyrus. Our results demonstrate that this method provides an efficient, interpretable, and generalizable framework for brain-age prediction, bridging the gap between CNN- and transformer-based approaches while opening new avenues for aging and neurodegeneration research.
Authors: David Yallup
Abstract: High-dimensional multimodal sampling problems from lattice field theory (LFT) have become important benchmarks for machine learning assisted sampling methods. We show that GPU-accelerated particle methods, Sequential Monte Carlo (SMC) and nested sampling, provide a strong classical baseline that matches or outperforms state-of-the-art neural samplers in sample quality and wall-clock time on standard scalar field theory benchmarks, while also estimating the partition function. Using only a single data-driven covariance for tuning, these methods achieve competitive performance without problem-specific structure, raising the bar for when learned proposals justify their training cost.
Authors: Jiajun Zhan, Zeyuan Ma, Yue-Jiao Gong, Kay Chen Tan
Abstract: Evolutionary multitasking (EMT) algorithms typically require tailored designs for knowledge transfer, in order to assure convergence and optimality in multitask optimization. In this paper, we explore designing a systematic and generalizable knowledge transfer policy through Reinforcement Learning. We first identify three major challenges: determining the task to transfer (where), the knowledge to be transferred (what) and the mechanism for the transfer (how). To address these challenges, we formulate a multi-role RL system where three (groups of) policy networks act as specialized agents: a task routing agent incorporates an attention-based similarity recognition module to determine source-target transfer pairs via attention scores; a knowledge control agent determines the proportion of elite solutions to transfer; and a group of strategy adaptation agents control transfer strength by dynamically controlling hyper-parameters in the underlying EMT framework. Through pre-training all network modules end-to-end over an augmented multitask problem distribution, a generalizable meta-policy is obtained. Comprehensive validation experiments show state-of-the-art performance of our method against representative baselines. Further in-depth analysis not only reveals the rationale behind our proposal but also provide insightful interpretations on what the system have learned.
Authors: Vladislav Pedashenko, Laida Kushnareva, Yana Khassan Nibal, Eduard Tulchinskii, Kristian Kuznetsov, Vladislav Zharchinskii, Yury Maximov, Irina Piontkovskaya
Abstract: Intrinsic dimension (ID) is an important tool in modern LLM analysis, informing studies of training dynamics, scaling behavior, and dataset structure, yet its textual determinants remain underexplored. We provide the first comprehensive study grounding ID in interpretable text properties through cross-encoder analysis, linguistic features, and sparse autoencoders (SAEs). In this work, we establish three key findings. First, ID is complementary to entropy-based metrics: after controlling for length, the two are uncorrelated, with ID capturing geometric complexity orthogonal to prediction quality. Second, ID exhibits robust genre stratification: scientific prose shows low ID (~8), encyclopedic content medium ID (~9), and creative/opinion writing high ID (~10.5) across all models tested. This reveals that contemporary LLMs find scientific text "representationally simple" while fiction requires additional degrees of freedom. Third, using SAEs, we identify causal features: scientific signals (formal tone, report templates, statistics) reduce ID; humanized signals (personalization, emotion, narrative) increase it. Steering experiments confirm these effects are causal. Thus, for contemporary models, scientific writing appears comparatively "easy", whereas fiction, opinion, and affect add representational degrees of freedom. Our multi-faceted analysis provides practical guidance for the proper use of ID and the sound interpretation of ID-based results.
Authors: Pol Ben\'itez, Cibr\'an L\'opez, Edgardo Saucedo, Teruyasu Mizoguchi, Claudio Cazorla
Abstract: Machine learning (ML) methods have become powerful tools for predicting material properties with near first-principles accuracy and vastly reduced computational cost. However, the performance of ML models critically depends on the quality, size, and diversity of the training dataset. In materials science, this dependence is particularly important for learning from low-symmetry atomistic configurations that capture thermal excitations, structural defects, and chemical disorder, features that are ubiquitous in real materials but underrepresented in most datasets. The absence of systematic strategies for generating representative training data may therefore limit the predictive power of ML models in technologically critical fields such as energy conversion and photonics. In this work, we assess the effectiveness of graph neural network (GNN) models trained on two fundamentally different types of datasets: one composed of randomly generated atomic configurations and another constructed using physically informed sampling based on lattice vibrations. As a case study, we address the challenging task of predicting electronic and mechanical properties of a prototypical family of optoelectronic materials under realistic finite-temperature conditions. We find that the phonons-informed model consistently outperforms the randomly trained counterpart, despite relying on fewer data points. Explainability analyses further reveal that high-performing models assign greater weight to chemically meaningful bonds that control property variations, underscoring the importance of physically guided data generation. Overall, this work demonstrates that larger datasets do not necessarily yield better GNN predictive models and introduces a simple and general strategy for efficiently constructing high-quality training data in materials informatics.
Authors: Tomas Espana, Yadh Hafsi, Fabrizio Lillo, Edoardo Vittori
Abstract: We investigate the use of Reinforcement Learning for the optimal execution of meta-orders, where the objective is to execute incrementally large orders while minimizing implementation shortfall and market impact over an extended period of time. Departing from traditional parametric approaches to price dynamics and impact modeling, we adopt a model-free, data-driven framework. Since policy optimization requires counterfactual feedback that historical data cannot provide, we employ the Queue-Reactive Model to generate realistic and tractable limit order book simulations that encompass transient price impact, and nonlinear and dynamic order flow responses. Methodologically, we train a Double Deep Q-Network agent on a state space comprising time, inventory, price, and depth variables, and evaluate its performance against established benchmarks. Numerical simulation results show that the agent learns a policy that is both strategic and tactical, adapting effectively to order book conditions and outperforming standard approaches across multiple training configurations. These findings provide strong evidence that model-free Reinforcement Learning can yield adaptive and robust solutions to the optimal execution problem.
Authors: Loveneet Saini, Hasan Tercan, Tobias Meisen
Abstract: Object detection with 3D radar is essential for 360-degree automotive perception, but radar's long wavelengths produce sparse and irregular reflections that challenge traditional grid and sequence-based convolutional and transformer detectors. This paper introduces Graph Query Networks (GQN), an attention-based framework that models objects sensed by radar as graphs, to extract individualized relational and contextual features. GQN employs a novel concept of graph queries to dynamically attend over the bird's-eye view (BEV) space, constructing object-specific graphs processed by two novel modules: EdgeFocus for relational reasoning and DeepContext Pooling for contextual aggregation. On the NuScenes dataset, GQN improves relative mAP by up to +53%, including a +8.2% gain over the strongest prior radar method, while reducing peak graph construction overhead by 80% with moderate FLOPs cost.
Authors: Abdelhamid Ezzerg, Ilija Bogunovic, Jeremias Knoblauch
Abstract: Bayesian Optimization is critically vulnerable to extreme outliers. Existing provably robust methods typically assume a bounded cumulative corruption budget, which makes them defenseless against even a single corruption of sufficient magnitude. To address this, we introduce a new adversary whose budget is only bounded in the frequency of corruptions, not in their magnitude. We then derive RCGP-UCB, an algorithm coupling the famous upper confidence bound (UCB) approach with a Robust Conjugate Gaussian Process (RCGP). We present stable and adaptive versions of RCGP-UCB, and prove that they achieve sublinear regret in the presence of up to $O(T^{1/2})$ and $O(T^{1/3})$ corruptions with possibly infinite magnitude. This robustness comes at near zero cost: without outliers, RCGP-UCB's regret bounds match those of the standard GP-UCB algorithm.
Authors: The Tien Mai
Abstract: In high-dimensional statistics, the Lasso is a cornerstone method for simultaneous variable selection and parameter estimation. However, its reliance on the squared loss function renders it highly sensitive to outliers and heavy-tailed noise, potentially leading to unreliable model selection and biased estimates. To address this limitation, we introduce the Exponential Lasso, a novel robust method that integrates an exponential-type loss function within the Lasso framework. This loss function is designed to achieve a smooth trade-off between statistical efficiency under Gaussian noise and robustness against data contamination. Unlike other methods that cap the influence of large residuals, the exponential loss smoothly redescends, effectively downweighting the impact of extreme outliers while preserving near-quadratic behavior for small errors. We establish theoretical guarantees showing that the Exponential Lasso achieves strong statistical convergence rates, matching the classical Lasso under ideal conditions while maintaining its robustness in the presence of heavy-tailed contamination. Computationally, the estimator is optimized efficiently via a Majorization-Minimization (MM) algorithm that iteratively solves a series of weighted Lasso subproblems. Numerical experiments demonstrate that the proposed method is highly competitive, outperforming the classical Lasso in contaminated settings and maintaining strong performance even under Gaussian noise. Our method is implemented in the \texttt{R} package \texttt{heavylasso} available on Github: https://github.com/tienmt/heavylasso
Authors: Spyridon Loukovitis, Vasileios Karampinis, Athanasios Voulodimos
Abstract: Developing reliable UAV navigation systems requires robust air-to-air object detectors capable of distinguishing between objects seen during training and previously unseen objects. While many methods address closed-set detection and achieve high-confidence recognition of in-domain (ID) targets, they generally do not tackle open-set detection, which requires simultaneous handling of both ID and out-of-distribution (OOD) objects. Existing open-set approaches typically rely on a single uncertainty score with thresholding, limiting flexibility and often conflating OOD objects with background clutter. In contrast, we propose a lightweight, model-agnostic post-processing framework that explicitly separates background from unknown objects while preserving the base detector's performance. Our approach extends open-set detection beyond binary ID/OOD classification to real-time three-way classification among ID targets, OOD objects, and background. To this end, we employ a fusion scheme that aggregates multiple confidence estimates and per-detection features using a compact multilayer perceptron (MLP). Incorporating different logit variants into the MLP consistently enhances performance across both binary and three-class classification without compromising throughput. Extensive ablation and comparative experiments confirm that our method surpasses threshold-based baselines in two-class classification by an average of 2.7% AUROC, while retaining or improving open-set mAP. Furthermore, our study uniquely enables robust three-class classification, a critical capability for safe UAV navigation, where OOD objects must be actively avoided and background regions safely ignored. Comparative analysis highlights that our method surpasses competitive techniques in AUROC across datasets, while improving closed-set mAP by up to 9 points, an 18% relative gain.
Authors: Luca Mossina, Corentin Friedrich
Abstract: Reliable semantic segmentation is essential for clinical decision making, yet deep models rarely provide explicit statistical guarantees on their errors. We introduce a simple post-hoc framework that constructs confidence masks with distribution-free, image-level control of false-positive predictions. Given any pretrained segmentation model, we define a nested family of shrunken masks obtained either by increasing the score threshold or by applying morphological erosion. A labeled calibration set is used to select a single shrink parameter via conformal prediction, ensuring that, for new images that are exchangeable with the calibration data, the proportion of false positives retained in the confidence mask stays below a user-specified tolerance with high probability. The method is model-agnostic, requires no retraining, and provides finite-sample guarantees regardless of the underlying predictor. Experiments on a polyp-segmentation benchmark demonstrate target-level empirical validity. Our framework enables practical, risk-aware segmentation in settings where over-segmentation can have clinical consequences. Code at https://github.com/deel-ai-papers/conseco.
Authors: Wenlun Zhang, Yunshan Zhong, Zihao Ding, Xinyu Li, Kentaro Yoshioka
Abstract: Data-Free Quantization (DFQ) offers a practical solution for model compression without requiring access to real data, making it particularly attractive in privacy-sensitive scenarios. While DFQ has shown promise for unimodal models, its extension to Vision-Language Models such as Contrastive Language-Image Pre-training (CLIP) models remains underexplored. In this work, we reveal that directly applying existing DFQ techniques to CLIP results in substantial performance degradation due to two key limitations: insufficient semantic content and low intra-image diversity in synthesized samples. To tackle these challenges, we propose D4C, the first DFQ framework tailored for CLIP. D4C synthesizes semantically rich and structurally diverse pseudo images through three key components: (1) Prompt-Guided Semantic Injection aligns generated images with real-world semantics using text prompts; (2) Structural Contrastive Generation reproduces compositional structures of natural images by leveraging foreground-background contrastive synthesis; and (3) Perturbation-Aware Enhancement applies controlled perturbations to improve sample diversity and robustness. These components jointly empower D4C to synthesize images that are both semantically informative and structurally diverse, effectively bridging the performance gap of DFQ on CLIP. Extensive experiments validate the effectiveness of D4C, showing significant performance improvements on various bit-widths and models. For example, under the W4A8 setting with CLIP ResNet-50 and ViT-B/32, D4C achieves Top-1 accuracy improvement of 12.4% and 18.9% on CIFAR-10, 6.8% and 19.7% on CIFAR-100, and 1.4% and 5.7% on ImageNet-1K in zero-shot classification, respectively.
Authors: Victorita Dolean, Daria Hrebenshchykova, St\'ephane Lanteri, Victor Michel-Dansac
Abstract: Accurately simulating wave propagation is crucial in fields such as acoustics, electromagnetism, and seismic analysis. Traditional numerical methods, like finite difference and finite element approaches, are widely used to solve governing partial differential equations (PDEs) such as the Helmholtz equation. However, these methods face significant computational challenges when applied to high-frequency wave problems in complex two-dimensional domains. This work investigates Finite Basis Physics-Informed Neural Networks (FBPINNs) and their multilevel extensions as a promising alternative. These methods leverage domain decomposition, partitioning the computational domain into overlapping sub-domains, each governed by a local neural network. We assess their accuracy and computational efficiency in solving the Helmholtz equation for the homogeneous case, demonstrating their potential to mitigate the limitations of traditional approaches.
Authors: Alexej Brauer, Mario V. W\"uthrich
Abstract: The Gini score is a popular tool in statistical modeling and machine learning for model validation and model selection. It is a purely rank based score that allows one to assess risk rankings. The Gini score for statistical modeling has mainly been used in a binary context, in which it has many equivalent reformulations such as the receiver operating characteristic (ROC) or the area under the curve (AUC). In the actuarial literature, this rank based score for binary responses has been extended to general real-valued random variables using Lorenz curves and concentration curves. While these initial concepts assume that the risk ranking is generated by a continuous distribution function, we discuss in this paper how the Gini score can be used in the case of ties in the risk ranking. Moreover, we adapt the Gini score to the common actuarial situation of having case weights.
Authors: Dabin Jeong, Amirhossein Vahidi, Ciro Ram\'irez-Su\'astegui, Marie Moullet, Kevin Ly, Mohammad Vali Sanian, Sebastian Birk, Yinshui Chang, Adam Boxall, Daniyal Jafree, Lloyd Steele, Vijaya Baskar MS, Muzlifah Haniffa, Mohammad Lotfollahi
Abstract: Recent advances in computational pathology have leveraged vision-language models to learn joint representations of Hematoxylin and Eosin (HE) images with spatial transcriptomic (ST) profiles. However, existing approaches typically align HE tiles with their corresponding ST profiles at a single scale, overlooking fine-grained cellular structures and their spatial organization. To address this, we propose Sigmma, a multi-modal contrastive alignment framework for learning hierarchical representations of HE images and spatial transcriptome profiles across multiple scales. Sigmma introduces multi-scale contrastive alignment, ensuring that representations learned at different scales remain coherent across modalities. Furthermore, by representing cell interactions as a graph and integrating inter- and intra-subgraph relationships, our approach effectively captures cell-cell interactions, ranging from fine to coarse, within the tissue microenvironment. We demonstrate that Sigmm learns representations that better capture cross-modal correspondences, leading to an improvement of avg. 9.78\% in the gene-expression prediction task and avg. 26.93\% in the cross-modal retrieval task across datasets. We further show that it learns meaningful multi-tissue organization in downstream analyses.
Authors: Rashid Iqbal, Saddam Hussain Khan
Abstract: This work proposes a hybrid deep learning approach, namely Residual and Spatial Learning based Channel Augmented Integrated CNN-Transformer architecture, that leverages the strengths of CNN and Transformer towards enhanced MPox detection. The proposed RS-CA-HSICT framework is composed of an HSICT block, a residual CNN module, a spatial CNN block, and a CA, which enhances the diverse feature space, detailed lesion information, and long-range dependencies. The new HSICT module first integrates an abstract representation of the stem CNN and customized ICT blocks for efficient multihead attention and structured CNN layers with homogeneous (H) and structural (S) operations. The customized ICT blocks learn global contextual interactions and local texture extraction. Additionally, H and S layers learn spatial homogeneity and fine structural details by reducing noise and modeling complex morphological variations. Moreover, inverse residual learning enhances vanishing gradient, and stage-wise resolution reduction ensures scale invariance. Furthermore, the RS-CA-HSICT framework augments the learned HSICT channels with the TL-driven Residual and Spatial CNN maps for enhanced multiscale feature space capturing global and localized structural cues, subtle texture, and contrast variations. These channels, preceding augmentation, are refined through the Channel-Fusion-and-Attention block, which preserves discriminative channels while suppressing redundant ones, thereby enabling efficient computation. Finally, the spatial attention mechanism refines pixel selection to detect subtle patterns and intra-class contrast variations in Mpox. Experimental results on both the Kaggle benchmark and a diverse MPox dataset reported classification accuracy as high as 98.30% and an F1-score of 98.13%, which outperforms the existing CNNs and ViTs.
Authors: Peiming Yang, Sankeerth Durvasula, Ivan Fernandez, Mohammad Sadrosadati, Onur Mutlu, Gennady Pekhimenko, Christina Giannoula
Abstract: Processing-In-Memory (PIM) devices integrated with high-performance Host processors (e.g., GPUs) can accelerate memory-intensive kernels in Machine Learning (ML) models, including Large Language Models (LLMs), by leveraging high memory bandwidth at PIM cores. However, Host processors and PIM cores require different data layouts: Hosts need consecutive elements distributed across DRAM banks, while PIM cores need them within local banks. This necessitates data rearrangements in ML kernel execution that pose significant performance and programmability challenges, further exacerbated by the need to support diverse PIM backends. Current compilation approaches lack systematic optimization for diverse ML kernels across multiple PIM backends and may largely ignore data rearrangements during compute code optimization. We demonstrate that data rearrangements and compute code optimization are interdependent, and need to be jointly optimized during the tuning process. To address this, we design DCC, the first data-centric ML compiler for PIM systems that jointly co-optimizes data rearrangements and compute code in a unified tuning process. DCC integrates a multi-layer PIM abstraction that enables various data distribution and processing strategies on different PIM backends. DCC enables effective co-optimization by mapping data partitioning strategies to compute loop partitions, applying PIM-specific code optimizations and leveraging a fast and accurate performance prediction model to select optimal configurations. Our evaluations in various individual ML kernels demonstrate that DCC achieves up to 7.68x speedup (2.7x average) on HBM-PIM and up to 13.17x speedup (5.75x average) on AttAcc PIM backend over GPU-only execution. In end-to-end LLM inference, DCC on AttAcc accelerates GPT-3 and LLaMA-2 by up to 7.71x (4.88x average) over GPU.
Authors: Yifei Gao, Hans J. He, Daniel J. Stilwell, James McMahon
Abstract: Teams of cooperating autonomous underwater vehicles (AUVs) rely on acoustic communication for coordination, yet this communication medium is constrained by limited range, multi-path effects, and low bandwidth. One way to address the uncertainty associated with acoustic communication is to learn the communication environment in real-time. We address the challenge of a team of robots building a map of the probability of communication success from one location to another in real-time. This is a decentralized classification problem -- communication events are either successful or unsuccessful -- where AUVs share a subset of their communication measurements to build the map. The main contribution of this work is a rigorously derived data sharing policy that selects measurements to be shared among AUVs. We experimentally validate our proposed sharing policy using real acoustic communication data collected from teams of Virginia Tech 690 AUVs, demonstrating its effectiveness in underwater environments.
Authors: Max Hirsch, Federico Pichi
Abstract: In multi-objective optimization, multiple loss terms are weighted and added together to form a single objective. These weights are chosen to properly balance the competing losses according to some meta-goal. For example, in physics-informed neural networks (PINNs), these weights are often adaptively chosen to improve the network's generalization error. A popular choice of adaptive weights is based on the neural tangent kernel (NTK) of the PINN, which describes the evolution of the network in predictor space during training. The convergence of such an adaptive weighting algorithm is not clear a priori. Moreover, these NTK-based weights would be updated frequently during training, further increasing the computational burden of the learning process. In this paper, we prove that under appropriate conditions, gradient descent enhanced with adaptive NTK-based weights is convergent in a suitable sense. We then address the problem of computational efficiency by developing a randomized algorithm inspired by a predictor-corrector approach and matrix sketching, which produces unbiased estimates of the NTK up to an arbitrarily small discretization error. Finally, we provide numerical experiments to support our theoretical findings and to show the efficacy of our randomized algorithm. Code Availability: https://github.com/maxhirsch/Efficient-NTK
Authors: Georgios Venianakis, Constantinos Theodoropoulos, Michail Kavousanakis
Abstract: Parameter estimation remains a challenging task across many areas of engineering. Because data acquisition can often be costly, limited, or prone to inaccuracies (noise, uncertainty) it is crucial to identify sensor configurations that provide the maximum amount of information about the unknown parameters, in particular for the case of distributed-parameter systems, where spatial variations are important. Physics-Informed Neural Networks (PINNs) have recently emerged as a powerful machine-learning (ML) tool for parameter estimation, particularly in cases with sparse or noisy measurements, overcoming some of the limitations of traditional optimization-based and Bayesian approaches. Despite the widespread use of PINNs for solving inverse problems, relatively little attention has been given to how their performance depends on sensor placement. This study addresses this gap by introducing a comprehensive PINN-based framework that simultaneously tackles optimal sensor placement and parameter estimation. Our approach involves training a PINN model in which the parameters of interest are included as additional inputs. This enables the efficient computation of sensitivity functions through automatic differentiation, which are then used to determine optimal sensor locations exploiting the D-optimality criterion. The framework is validated on two illustrative distributed-parameter reaction-diffusion-advection problems of increasing complexity. The results demonstrate that our PINNs-based methodology consistently achieves higher accuracy compared to parameter values estimated from intuitively or randomly selected sensor positions.
Authors: Miruna-Alexandra Gafencu, Yordanka Velikova, Nassir Navab, Mohammad Farid Azampour
Abstract: Ultrasound offers a radiation-free, cost-effective solution for real-time visualization of spinal landmarks, paraspinal soft tissues and neurovascular structures, making it valuable for intraoperative guidance during spinal procedures. However, ultrasound suffers from inherent limitations in visualizing complete vertebral anatomy, in particular vertebral bodies, due to acoustic shadowing effects caused by bone. In this work, we present a novel multi-modal deep learning method for completing occluded anatomical structures in 3D ultrasound by leveraging complementary information from a single X-ray image. To enable training, we generate paired training data consisting of: (1) 2D lateral vertebral views that simulate X-ray scans, and (2) 3D partial vertebrae representations that mimic the limited visibility and occlusions encountered during ultrasound spine imaging. Our method integrates morphological information from both imaging modalities and demonstrates significant improvements in vertebral reconstruction (p < 0.001) compared to state of art in 3D ultrasound vertebral completion. We perform phantom studies as an initial step to future clinical translation, and achieve a more accurate, complete volumetric lumbar spine visualization overlayed on the ultrasound scan without the need for registration with preoperative modalities such as computed tomography. This demonstrates that integrating a single X-ray projection mitigates ultrasound's key limitation while preserving its strengths as the primary imaging modality. Code and data can be found at https://github.com/miruna20/US-X-Complete
Authors: G\'abor Bal\'azs
Abstract: This paper presents a tractable algorithm for estimating an unknown Lipschitz function from noisy observations and establishes an upper bound on its convergence rate. The approach extends max-affine methods from convex shape-restricted regression to the more general Lipschitz setting. A key component is a nonlinear feature expansion that maps max-affine functions into a subclass of delta-convex functions, which act as universal approximators of Lipschitz functions while preserving their Lipschitz constants. Leveraging this property, the estimator attains the minimax convergence rate (up to logarithmic factors) with respect to the intrinsic dimension of the data under squared loss and subgaussian distributions in the random design setting. The algorithm integrates adaptive partitioning to capture intrinsic dimension, a penalty-based regularization mechanism that removes the need to know the true Lipschitz constant, and a two-stage optimization procedure combining a convex initialization with local refinement. The framework is also straightforward to adapt to convex shape-restricted regression. Experiments demonstrate competitive performance relative to other theoretically justified methods, including nearest-neighbor and kernel-based regressors.
Authors: Petrus E. O. G. B. Abreu, Gabriela M. M. Paix\~ao, Jiawei Li, Paulo R. Gomes, Peter W. Macfarlane, Ana C. S. Oliveira, Vinicius T. Carvalho, Thomas B. Sch\"on, Antonio Luiz P. Ribeiro, Ant\^onio H. Ribeiro
Abstract: Data-driven methods for electrocardiogram (ECG) interpretation are rapidly progressing. Large datasets have enabled advances in artificial intelligence (AI) based ECG analysis, yet limitations in annotation quality, size, and scope remain major challenges. Here we present CODE-II, a large-scale real-world dataset of 2,735,269 12-lead ECGs from 2,093,807 adult patients collected by the Telehealth Network of Minas Gerais (TNMG), Brazil. Each exam was annotated using standardized diagnostic criteria and reviewed by cardiologists. A defining feature of CODE-II is a set of 66 clinically meaningful diagnostic classes, developed with cardiologist input and routinely used in telehealth practice. We additionally provide an open available subset: CODE-II-open, a public subset of 15,000 patients, and the CODE-II-test, a non-overlapping set of 8,475 exams reviewed by multiple cardiologists for blinded evaluation. A neural network pre-trained on CODE-II achieved superior transfer performance on external benchmarks (PTB-XL and CPSC 2018) and outperformed alternatives trained on larger datasets.
Authors: Tao Hu, Lan Li, Zhen-Hao Xie, Da-Wei Zhou
Abstract: Class-Incremental Learning (CIL) enables models to learn new classes continually while preserving past knowledge. Recently, vision-language models like CLIP offer transferable features via multi-modal pre-training, making them well-suited for CIL. However, real-world visual and linguistic concepts are inherently hierarchical: a textual concept like "dog" subsumes fine-grained categories such as "Labrador" and "Golden Retriever," and each category entails its images. But existing CLIP-based CIL methods fail to explicitly capture this inherent hierarchy, leading to fine-grained class features drift during incremental updates and ultimately to catastrophic forgetting. To address this challenge, we propose HASTEN (Hierarchical Semantic Tree Anchoring) that anchors hierarchical information into CIL to reduce catastrophic forgetting. First, we employ an external knowledge graph as supervision to embed visual and textual features in hyperbolic space, effectively preserving hierarchical structure as data evolves. Second, to mitigate catastrophic forgetting, we project gradients onto the null space of the shared hyperbolic mapper, preventing interference with prior tasks. These two steps work synergistically to enable the model to resist forgetting by maintaining hierarchical relationships. Extensive experiments show that HASTEN consistently outperforms existing methods while providing a unified structured representation.
Authors: Benjamin Dupuis, Mert G\"urb\"uzbalaban, Umut \c{S}im\c{s}ekli, Jian Wang, Sinan Yildirim, Lingjiong Zhu
Abstract: Characterizing the differential privacy (DP) of learning algorithms has become a major challenge in recent years. In parallel, many studies suggested investigating the behavior of stochastic gradient descent (SGD) with heavy-tailed noise, both as a model for modern deep learning models and to improve their performance. However, most DP bounds focus on light-tailed noise, where satisfactory guarantees have been obtained but the proposed techniques do not directly extend to the heavy-tailed setting. Recently, the first DP guarantees for heavy-tailed SGD were obtained. These results provide $(0,\delta)$-DP guarantees without requiring gradient clipping. Despite casting new light on the link between DP and heavy-tailed algorithms, these results have a strong dependence on the number of parameters and cannot be extended to other DP notions like the well-established R\'enyi differential privacy (RDP). In this work, we propose to address these limitations by deriving the first RDP guarantees for heavy-tailed SDEs, as well as their discretized counterparts. Our framework is based on new R\'enyi flow computations and the use of well-established fractional Poincar\'e inequalities. Under the assumption that such inequalities are satisfied, we obtain DP guarantees that have a much weaker dependence on the dimension compared to prior art.
Authors: Yicheng He, Chengsong Huang, Zongxia Li, Jiaxin Huang, Yonghui Yang
Abstract: Reinforcement learning (RL) provides a principled framework for improving Vision-Language Models (VLMs) on complex reasoning tasks. However, existing RL approaches often rely on human-annotated labels or task-specific heuristics to define verifiable rewards, both of which are costly and difficult to scale. We introduce VisPlay, a self-evolving RL framework that enables VLMs to autonomously improve their reasoning abilities using large amounts of unlabeled image data. Starting from a single base VLM, VisPlay assigns the model into two interacting roles: an Image-Conditioned Questioner that formulates challenging yet answerable visual questions, and a Multimodal Reasoner that generates silver responses. These roles are jointly trained with Group Relative Policy Optimization (GRPO), which incorporates diversity and difficulty rewards to balance the complexity of generated questions with the quality of the silver answers. VisPlay scales efficiently across two model families. When trained on Qwen2.5-VL and MiMo-VL, VisPlay achieves consistent improvements in visual reasoning, compositional generalization, and hallucination reduction across eight benchmarks, including MM-Vet and MMMU, demonstrating a scalable path toward self-evolving multimodal intelligence. The project page is available at https://bruno686.github.io/VisPlay/
Authors: Jianqiao Mao, Max A. Little
Abstract: Front-door adjustment provides a simple closed-form identification formula under the classical front-door criterion, but its applicability is often viewed as narrow and strict. Although ID algorithm is very useful and is proved effective for causal relation identification in general causal graphs (if it is identifiable), performing ID algorithm does not guarantee to obtain a practical, easy-to-estimate interventional distribution expression. We argue that the applicability of the front-door criterion is not as limited as it seems: many more complicated causal graphs can be reduced to the front-door criterion. In this paper, We introduce front-door reducibility (FDR), a graphical condition on acyclic directed mixed graphs (ADMGs) that extends the applicability of the classic front-door criterion to reduce a large family of complicated causal graphs to a front-door setting by aggregating variables into super-nodes (FDR triple) $\left(\boldsymbol{X}^{*},\boldsymbol{Y}^{*},\boldsymbol{M}^{*}\right)$. After characterizing FDR criterion, we prove a graph-level equivalence between the satisfication of FDR criterion and the applicability of FDR adjustment. Meanwhile, we then present FDR-TID, an exact algorithm that detects an admissible FDR triple, together with established the algorithm's correctness, completeness, and finite termination. Empirically-motivated examples illustrate that many graphs outside the textbook front-door setting are FDR, yielding simple, estimable adjustments where general ID expressions would be cumbersome. FDR thus complements existing identification method by prioritizing interpretability and computational simplicity without sacrificing generality across mixed graphs.
Authors: Naveen Raman, Jingwu Tang, Zhiyu Chen, Zheyuan Ryan Shi, Sean Hudson, Ameesh Kapoor, Fei Fang
Abstract: Food rescue organizations simultaneously tackle food insecurity and waste by working with volunteers to redistribute food from donors who have excess to recipients who need it. Volunteer feedback allows food rescue organizations to identify issues early and ensure volunteer satisfaction. However, food rescue organizations monitor feedback manually, which can be cumbersome and labor-intensive, making it difficult to prioritize which issues are most important. In this work, we investigate how large language models (LLMs) assist food rescue organizers in understanding and taking action based on volunteer experiences. We work with 412 Food Rescue, a large food rescue organization based in Pittsburgh, Pennsylvania, to design RescueLens, an LLM-powered tool that automatically categorizes volunteer feedback, suggests donors and recipients to follow up with, and updates volunteer directions based on feedback. We evaluate the performance of RescueLens on an annotated dataset, and show that it can recover 96% of volunteer issues at 71% precision. Moreover, by ranking donors and recipients according to their rates of volunteer issues, RescueLens allows organizers to focus on 0.5% of donors responsible for more than 30% of volunteer issues. RescueLens is now deployed at 412 Food Rescue and through semi-structured interviews with organizers, we find that RescueLens streamlines the feedback process so organizers better allocate their time.
Authors: Violeta Kastreva, Philip Whittington, Dennis Komm, Tiago Pimentel
Abstract: Recent works have shown that tokenisation is NP-complete. However, these works assume tokenisation is applied to inputs with unboundedly large alphabets -- an unrealistic assumption, given that in practice tokenisers operate over fixed-size alphabets, such as bytes or Unicode characters. We close this gap by analysing tokenisation over bounded $n$-ary alphabets, considering two natural variants: bottom-up tokenisation and direct tokenisation, where we must, respectively, select a sequence of merge operations or a vocabulary whose application optimally compresses a dataset. First, we note that proving hardness results for an $n$-ary alphabet proves the same results for alphabets of any larger size. We then prove that even with binary alphabets, both variants are not only NP-complete, but admit no polynomial-time approximation scheme (unless P=NP). We further show that direct tokenisation remains NP-complete even when applied to unary alphabets. While unary alphabets may not be practically useful, this result establishes that the computational intractability of tokenisation is not an artifact of large alphabets or complex constructions, but a fundamental barrier. Overall, our results explain why practical algorithms such as BPE and UnigramLM are heuristic, and points toward approximation algorithms being an important path going forward for tokenisation research.
Authors: Shahryar Zehtabi, Seyyedali Hosseinalipour, Christopher G. Brinton
Abstract: Federated learning (FL) is a popular technique for distributing machine learning (ML) across a set of edge devices. In this paper, we study fully decentralized FL, where in addition to devices conducting training locally, they carry out model aggregations via cooperative consensus formation over device-to-device (D2D) networks. We introduce asynchronous, event-triggered communications among the devices to handle settings where access to a central server is not feasible. To account for the inherent resource heterogeneity and statistical diversity challenges in FL, we define personalized communication triggering conditions at each device that weigh the change in local model parameters against the available local network resources. We theoretically recover the $O(\ln{k} / \sqrt{k})$ convergence rate to the globally optimal model of decentralized gradient descent (DGD) methods in the setup of our methodology. We provide our convergence guarantees for the last iterates of models, under relaxed graph connectivity and data heterogeneity assumptions compared with the existing literature. To do so, we demonstrate a $B$-connected information flow guarantee in the presence of sporadic communications over the time-varying D2D graph. Our subsequent numerical evaluations demonstrate that our methodology obtains substantial improvements in convergence speed and/or communication savings compared to existing decentralized FL baselines.
Authors: Wenjian Hao, Paulo C. Heredia, Shaoshuai Mou
Abstract: This paper presents a data-driven method to find a closed-loop optimal controller, which minimizes a specified infinite-horizon cost function for systems with unknown dynamics. Suppose the closed-loop optimal controller can be parameterized by a given class of functions, hereafter referred to as the policy. The proposed method introduces a novel gradient estimation framework, which approximates the gradient of the cost function with respect to the policy parameters via integrating the Koopman operator with the classical concept of actor-critic. This enables the policy parameters to be tuned iteratively using gradient descent to achieve an optimal controller, leveraging the linearity of the Koopman operator. The convergence analysis of the proposed framework is provided. The control performance of the proposed method is evaluated through simulations compared with classical optimal control methods that usually assume the dynamics are known.
Authors: Quang-Huy Nguyen, Jin Peng Zhou, Zhenzhen Liu, Khanh-Huyen Bui, Kilian Q. Weinberger, Wei-Lun Chao, Dung D. Le
Abstract: Recent object detectors have achieved impressive accuracy in identifying objects seen during training. However, real-world deployment often introduces novel and unexpected objects, referred to as out-of-distribution (OOD) objects, posing significant challenges to model trustworthiness. Modern object detectors are typically overconfident, making it unreliable to use their predictions alone for OOD detection. To address this, we propose leveraging an auxiliary model as a complementary solution. Specifically, we utilize an off-the-shelf text-to-image generative model, such as Stable Diffusion, which is trained with objective functions distinct from those of discriminative object detectors. We hypothesize that this fundamental difference enables the detection of OOD objects by measuring inconsistencies between the models. Concretely, for a given detected object bounding box and its predicted in-distribution class label, we perform class-conditioned inpainting on the image with the object removed. If the object is OOD, the inpainted image is likely to deviate significantly from the original, making the reconstruction error a robust indicator of OOD status. Extensive experiments demonstrate that our approach consistently surpasses existing zero-shot and non-zero-shot OOD detection methods, establishing a robust framework for enhancing object detection systems in dynamic environments.
Authors: Guner Dilsad Er, Sebastian Trimpe, Michael Muehlebach
Abstract: We consider a distributed learning problem, where agents minimize a global objective function by exchanging information over a network. Our approach has two distinct features: (i) It substantially reduces communication by triggering communication only when necessary, and (ii) it is agnostic to the data-distribution among the different agents. We therefore guarantee convergence even if the local data-distributions of the agents are arbitrarily distinct. We analyze the convergence rate of the algorithm both in convex and nonconvex settings and derive accelerated convergence rates for the convex case. We also characterize the effect of communication failures and demonstrate that our algorithm is robust to these. The article concludes by presenting numerical results from distributed learning tasks on the MNIST and CIFAR-10 datasets. The experiments underline communication savings of 35% or more due to the event-based communication strategy, show resilience towards heterogeneous data-distributions, and highlight that our approach outperforms common baselines such as FedAvg, FedProx, SCAFFOLD and FedADMM.
Authors: Juan Miguel Lopez Alcaraz, Nils Strodthoff
Abstract: Despite the excelling performance of machine learning models, understanding their decisions remains a long-standing goal. Although commonly used attribution methods from explainable AI attempt to address this issue, they typically rely on associational rather than causal relationships. In this study, within the context of time series classification, we introduce a novel model-agnostic attribution method to assess the causal effect of concepts i.e., predefined segments within a time series, on classification outcomes. Our approach compares these causal attributions with closely related associational attributions, both theoretically and empirically. To estimate counterfactual outcomes, we use state-of-the-art diffusion models backed by state space models. We demonstrate the insights gained by our approach for a diverse set of qualitatively different time series classification tasks. Although causal and associational attributions might often share some similarities, in all cases they differ in important details, underscoring the risks associated with drawing causal conclusions from associational data alone. We believe that the proposed approach is also widely applicable in other domains to shed some light on the limits of associational attributions.
Authors: Faried Abu Zaid, Daniel Neider, Mustafa Yal\c{c}{\i}ner
Abstract: Formal verification has emerged as a promising method to ensure the safety and reliability of neural networks. However, many relevant properties, such as fairness or global robustness, pertain to the entire input space. If one applies verification techniques naively, the neural network is checked even on inputs that do not occur in the real world and have no meaning. To tackle this shortcoming, we propose the VeriFlow architecture as a flow-based density model tailored to allow any verification approach to restrict its search to some data distribution of interest. We argue that our architecture is particularly well suited for this purpose because of two major properties. First, we show that the transformation that is defined by our model is piecewise affine. Therefore, the model allows the usage of verifiers based on constraint solving with linear arithmetic. Second, upper density level sets (UDL) of the data distribution are definable via linear constraints in the latent space. As a consequence, representations of UDLs specified by a given probability are effectively computable in the latent space. This property allows for effective verification with a fine-grained, probabilistically interpretable control of how a-typical the inputs subject to verification are.
Authors: Pavel Rytir, Ales Wodecki, Jakub Marecek
Abstract: There has been a growing interest in causal learning in recent years. Commonly used representations of causal structures, including Bayesian networks and structural equation models (SEM), take the form of directed acyclic graphs (DAGs). We provide a novel mixed-integer quadratic programming formulation and an associated algorithm that identifies DAGs with a low structural Hamming distance between the identified DAG and the ground truth, under identifiability assumptions. The eventual exact learning is guaranteed by the global convergence of the branch-and-bound-and-cut algorithm, which is utilized. In addition to this, integer programming techniques give us access to the dual bound, which allows for a real time assessment of the quality of solution. Previously, integer programming techniques have been shown to lead to limited scaling in the case of DAG identification due to the super exponential number of constraints, which prevent the formation of cycles. The algorithm proposed circumvents this by selectively generating only the violated constraints using the so-called "lazy" constraints methodology. Our empirical results show that ExDAG outperforms state-of-the-art solvers in terms of structural Hamming distance and $F_1$ score when considering Gaussian noise on medium-sized graphs.
Authors: Keyu Chen, Ziqian Bi, Qian Niu, Junyu Liu, Benji Peng, Sen Zhang, Ming Liu, Xinyuan Song, Zekun Jiang, Tianyang Wang, Ming Li, Xuanhe Pan, Jiawei Xu, Jinlang Wang, Pohsun Feng
Abstract: The application of TensorFlow pre-trained models in deep learning is explored, with an emphasis on practical guidance for tasks such as image classification and object detection. The study covers modern architectures, including ResNet, MobileNet, and EfficientNet, and demonstrates the effectiveness of transfer learning through real-world examples and experiments. A comparison of linear probing and model fine-tuning is presented, supplemented by visualizations using techniques like PCA, t-SNE, and UMAP, allowing for an intuitive understanding of the impact of these approaches. The work provides complete example code and step-by-step instructions, offering valuable insights for both beginners and advanced users. By integrating theoretical concepts with hands-on practice, the paper equips readers with the tools necessary to address deep learning challenges efficiently.
Authors: Yifeng Ding, Hantian Ding, Shiqi Wang, Qing Sun, Varun Kumar, Zijian Wang
Abstract: Fill-in-the-Middle (FIM), or infilling, has become integral to code language models, enabling generation of missing code given both left and right contexts. However, the current FIM training paradigm which performs next-token prediction (NTP) over reordered sequence often leads to models struggling to generate content that aligns well with the surrounding context. We hypothesize that NTP alone is insufficient for models to learn effective planning conditioned on the distant right context, a critical factor for successful code infilling. To overcome this, we propose Horizon-Length Prediction (HLP), a novel training objective that teaches models to predict the number of remaining middle tokens at each step. HLP advances FIM with lookahead planning, enabling models to inherently learn infilling boundaries for arbitrary left and right contexts without relying on dataset-specific post-processing. Our evaluation across different model families and sizes shows that HLP significantly improves FIM performance by up to 24% relatively on diverse benchmarks, across file-level and repository-level. Furthermore, the enhanced planning capability gained through HLP boosts model performance on code reasoning. Importantly, HLP incurs negligible training overhead and no additional inference cost, ensuring its practicality for real-world scenarios.
Authors: Tao Sun, Xinwang Liu, Kun Yuan
Abstract: Gradient clipping has long been considered essential for ensuring the convergence of Stochastic Gradient Descent (SGD) in the presence of heavy-tailed gradient noise. In this paper, we revisit this belief and explore whether gradient normalization can serve as an effective alternative or complement. We prove that, under individual smoothness assumptions, gradient normalization alone is sufficient to guarantee convergence of the nonconvex SGD. Moreover, when combined with clipping, it yields far better rates of convergence under more challenging noise distributions. We provide a unifying theory describing normalization-only, clipping-only, and combined approaches. Moving forward, we investigate existing variance-reduced algorithms, establishing that, in such a setting, normalization alone is sufficient for convergence. Finally, we present an accelerated variant that under second-order smoothness improves convergence. Our results provide theoretical insights and practical guidance for using normalization and clipping in nonconvex optimization with heavy-tailed noise.
Authors: Maurice Kraus, Felix Divo, Devendra Singh Dhami, Kristian Kersting
Abstract: Time series data is prevalent across numerous fields, necessitating the development of robust and accurate forecasting models. Capturing patterns both within and between temporal and multivariate components is crucial for reliable predictions. We introduce xLSTM-Mixer, a model designed to effectively integrate temporal sequences, joint time-variate information, and multiple perspectives for robust forecasting. Our approach begins with a linear forecast shared across variates, which is then refined by xLSTM blocks. They serve as key elements for modeling the complex dynamics of challenging time series data. xLSTM-Mixer ultimately reconciles two distinct views to produce the final forecast. Our extensive evaluations demonstrate its superior long-term forecasting performance compared to recent state-of-the-art methods while requiring very little memory. A thorough model analysis provides further insights into its key components and confirms its robustness and effectiveness. This work contributes to the resurgence of recurrent models in forecasting by combining them, for the first time, with mixing architectures.
Authors: Adib Karimi, Mohammad Mehdi Ebadzadeh
Abstract: We propose a novel Inverse Reinforcement Learning (IRL) method that mitigates the rigidity of fixed reward structures and the limited flexibility of implicit reward regularization. Building on the Maximum Entropy IRL framework, our approach incorporates a squared temporal-difference (TD) regularizer with adaptive targets that evolve dynamically during training, thereby imposing adaptive bounds on recovered rewards and promoting robust decision-making. To capture richer return information, we integrate distributional RL into the learning process. Empirically, our method achieves expert-level performance on complex MuJoCo and Adroit environments, surpassing baseline methods on the Humanoid-v2 task with limited expert demonstrations. Extensive experiments and ablation studies further validate the effectiveness of the approach and provide insights into reward dynamics in imitation learning. Our source code is available at https://github.com/adibka/RIZE.
Authors: Evgeniia Vu, Andrei Boiarov, Dmitry Vetrov
Abstract: Generating co-speech gestures in real time requires both temporal coherence and efficient sampling. We introduce a novel framework for streaming gesture generation that extends Rolling Diffusion models with structured progressive noise scheduling, enabling seamless long-sequence motion synthesis while preserving realism and diversity. Our framework is universally compatible with existing diffusion-based gesture generation model, transforming them into streaming methods capable of continuous generation without requiring post-processing. We evaluate our framework on ZEGGS and BEAT, strong benchmarks for real-world applicability. Applied to state-of-the-art baselines on both datasets, it consistently outperforms them, demonstrating its effectiveness as a generalizable and efficient solution for real-time co-speech gesture synthesis. We further propose Rolling Diffusion Ladder Acceleration (RDLA), a new approach that employs a ladder-based noise scheduling strategy to simultaneously denoise multiple frames. This significantly improves sampling efficiency while maintaining motion consistency, achieving up to a 4x speedup with high visual fidelity and temporal coherence in our experiments. Comprehensive user studies further validate our framework ability to generate realistic, diverse gestures closely synchronized with the audio input.
Authors: Shubham Kumar, Narendra Ahuja
Abstract: Deep vision models perform input-output computations that are hard to interpret. Concept-based explanation methods (CBEMs) increase interpretability by re-expressing parts of the model with human-understandable semantic units, or concepts. Checking if the derived explanations are faithful -- that is, they represent the model's internal computation -- requires a surrogate that combines concepts to compute the output. Simplifications made for interpretability inevitably reduce faithfulness, resulting in a tradeoff between the two. State-of-the-art unsupervised CBEMs (U-CBEMs) have reported increasingly interpretable concepts, while also being more faithful to the model. However, we observe that the reported improvement in faithfulness artificially results from either (1) using overly complex surrogates, which introduces an unmeasured cost to the explanation's interpretability, or (2) relying on deletion-based approaches that, as we demonstrate, do not properly measure faithfulness. We propose Surrogate Faithfulness (SURF), which (1) replaces prior complex surrogates with a simple, linear surrogate that measures faithfulness without changing the explanation's interpretability and (2) introduces well-motivated metrics that assess loss across all output classes, not just the predicted class. We validate SURF with a measure-over-measure study by proposing a simple sanity check -- explanations with random concepts should be less faithful -- which prior surrogates fail. SURF enables the first reliable faithfulness benchmark of U-CBEMs, revealing that many visually compelling U-CBEMs are not faithful. Code to be released.
Authors: Akira Tamamori
Abstract: Kernel-based learning methods such as Kernel Logistic Regression (KLR) can dramatically increase the storage capacity of Hopfield networks, but the principles governing their performance and stability remain largely uncharacterized. This paper presents a comprehensive quantitative analysis of the attractor landscape in KLR-trained networks to establish a solid foundation for their design and application. Through extensive, statistically validated simulations, we address critical questions of generality, scalability, and robustness. Our comparative analysis reveals that KLR and Kernel Ridge Regression (KRR) exhibit similarly high storage capacities and clean attractor landscapes, suggesting this is a general property of kernel regression methods, though KRR is computationally much faster. We uncover a non-trivial, scale-dependent scaling law for the kernel width ($\gamma$), demonstrating that optimal capacity requires gamma to be scaled such that $\gamma \times N$ increases with network size $N$. This implies that larger networks necessitate more localized kernels -- where each pattern's influence is more spatially confined--to manage inter-pattern interference. Under this optimized scaling, we provide definitive evidence that the storage capacity scales linearly with network size ($P \propto N$). Furthermore, our sensitivity analysis shows that performance is remarkably robust to the choice of the regularization parameter lambda. Collectively, these findings provide a clear set of empirical principles for designing high-capacity, robust associative memories and clarify the mechanisms that enable kernel methods to overcome the classical limitations of Hopfield-type models.
Authors: Nikolaos Louloudakis, Ajitha Rajan
Abstract: With over 760 stars on GitHub and being part of the official ONNX repository, the ONNX Optimizer is the default tool for applying graph-based optimizations to ONNX models. Despite its widespread use, its ability to maintain model accuracy during optimization has not been thoroughly investigated. In this work, we present OODTE, a utility designed to automatically and comprehensively evaluate the correctness of the ONNX Optimizer. OODTE adopts a straightforward yet powerful differential testing and evaluation methodology, which can be readily adapted for use with other compiler optimizers. Specifically, OODTE takes a collection of ONNX models, applies optimizations, and executes both the original and optimized versions across a user-defined input set, automatically capturing any issues encountered during optimization. When discrepancies in accuracy arise, OODTE iteratively isolates the responsible optimization pass by repeating the process at a finer granularity. We applied OODTE to 130 well-known models from the official ONNX Model Hub, spanning diverse tasks including classification, object detection, semantic segmentation, text summarization, question answering, and sentiment analysis. Our evaluation revealed that 9.2% of the model instances either caused the optimizer to crash or led to the generation of invalid models using default optimization strategies. Additionally, 30% of classification models and 16.6% of object detection and segmentation models exhibited differing outputs across original and optimized versions, whereas models focused on text-related tasks were generally robust to optimization. OODTE uncovered 15 issues-14 previously unknown-affecting 9 of 47 optimization passes and the optimizer overall. All issues were reported to the ONNX Optimizer team. OODTE offers a simple but effective framework for validating AI model optimizers, applicable beyond the ONNX ecosystem.
Authors: Amir Rezaei Balef, Claire Vernade, Katharina Eggensperger
Abstract: The Combined Algorithm Selection and Hyperparameter optimization (CASH) is a challenging resource allocation problem in the field of AutoML. We propose MaxUCB, a max k-armed bandit method to trade off exploring different model classes and conducting hyperparameter optimization. MaxUCB is specifically designed for the light-tailed and bounded reward distributions arising in this setting and, thus, provides an efficient alternative compared to classic max k-armed bandit methods assuming heavy-tailed reward distributions. We theoretically and empirically evaluate our method on four standard AutoML benchmarks, demonstrating superior performance over prior approaches. We make our code and data available at https://github.com/amirbalef/CASH_with_Bandits
Authors: Shun Wang, Shun-Li Shang, Zi-Kui Liu, Wenrui Hao
Abstract: Traditional entropy-based methods - such as cross-entropy loss in classification problems - have long been essential tools for representing the information uncertainty and physical disorder in data and for developing artificial intelligence algorithms. However, the rapid growth of data across various domains has introduced new challenges, particularly the integration of heterogeneous datasets with intrinsic disparities. To address this, we introduce a zentropy-enhanced neural network (ZENN), extending zentropy theory into the data science domain via intrinsic entropy, enabling more effective learning from heterogeneous data sources. ZENN simultaneously learns both energy and intrinsic entropy components, capturing the underlying structure of multi-source data. To support this, we redesign the neural network architecture to better reflect the intrinsic properties and variability inherent in diverse datasets. We demonstrate the effectiveness of ZENN on classification tasks and energy landscape reconstructions, showing its superior generalization capabilities and robustness-particularly in predicting high-order derivatives. ZENN demonstrates superior generalization by introducing a learnable temperature variable that models latent multi-source heterogeneity, allowing it to surpass state-of-the-art models on CIFAR-10/100, BBCNews, and AGNews. As a practical application in materials science, we employ ZENN to reconstruct the Helmholtz energy landscape of Fe$_3$Pt using data generated from density functional theory (DFT) and capture key material behaviors, including negative thermal expansion and the critical point in the temperature-pressure space. Overall, this work presents a zentropy-grounded framework for data-driven machine learning, positioning ZENN as a versatile and robust approach for scientific problems involving complex, heterogeneous datasets.
Authors: Dibyajyoti Nayak, Somdatta Goswami
Abstract: Accurate temporal extrapolation remains a fundamental challenge for neural operators modeling dynamical systems, where predictions must extend far beyond the training horizon. Conventional DeepONet approaches rely on two limited paradigms: fixed-horizon rollouts, which predict full spatiotemporal solutions while ignoring temporal causality, and autoregressive schemes, which accumulate errors through sequential prediction. We introduce TI-DeepONet, a framework that integrates neural operators with adaptive numerical time-stepping to preserve the Markovian structure of dynamical systems while mitigating long-term error growth. Our method shifts the learning objective from direct state prediction to approximating instantaneous time-derivative fields, which are then integrated using standard numerical solvers. This naturally enables continuous-time prediction and allows the use of higher-order integrators at inference than those used in training, improving both efficiency and accuracy. We further propose TI(L)-DeepONet, which incorporates learnable coefficients for intermediate slopes in multi-stage integration, adapting to solution-specific dynamics and enhancing fidelity. Across four canonical PDEs featuring chaotic, dissipative, dispersive, and high-dimensional behavior, TI(L)-DeepONet slightly outperforms TI-DeepONet, and both achieve major reductions in relative L2 extrapolation error: about 81% compared to autoregressive methods and 70% compared to fixed-horizon approaches. Notably, both models maintain stable predictions over temporal domains nearly twice the training interval. This work establishes a physics-aware operator learning framework that bridges neural approximation with numerical analysis principles, addressing a key gap in long-term forecasting of complex physical systems.
Authors: Hao Wu, Yuan Gao, Chang Liu, Fan Xu, Fan Zhang, Zhihong Zhu, Yuqi Li, Xian Wu, Yuxuan Liang, Li Liu, Qingsong Wen, Kun Wang, Yu Zheng, Xiaomeng Huang
Abstract: Accurately predicting the long-term evolution of turbulence is crucial for advancing scientific understanding and optimizing engineering applications. However, existing deep learning methods face significant bottlenecks in long-term autoregressive prediction, which exhibit excessive smoothing and fail to accurately track complex fluid dynamics. Our extensive experimental and spectral analysis of prevailing methods provides an interpretable explanation for this shortcoming, identifying Spectral Bias as the core obstacle. Concretely, spectral bias is the inherent tendency of models to favor low-frequency, smooth features while overlooking critical high-frequency details during training, thus reducing fidelity and causing physical distortions in long-term predictions. Building on this insight, we propose Turb-L1, an innovative turbulence prediction method, which utilizes a Hierarchical Dynamics Synthesis mechanism within a multi-grid architecture to explicitly overcome spectral bias. It accurately captures cross-scale interactions and preserves the fidelity of high-frequency dynamics, enabling reliable long-term tracking of turbulence evolution. Extensive experiments on the 2D turbulence benchmark show that Turb-L1 demonstrates excellent performance: (I) In long-term predictions, it reduces Mean Squared Error (MSE) by $80.3\%$ and increases Structural Similarity (SSIM) by over $9\times$ compared to the SOTA baseline, significantly improving prediction fidelity. (II) It effectively overcomes spectral bias, accurately reproducing the full enstrophy spectrum and maintaining physical realism in high-wavenumber regions, thus avoiding the spectral distortions or spurious energy accumulation seen in other methods.
Authors: Dongyeop Woo, Minsu Kim, Minkyu Kim, Kiyoung Seong, Sungsoo Ahn
Abstract: We propose Energy-based generator matching (EGM), a modality-agnostic approach to train generative models from energy functions in the absence of data. Extending the recently proposed generator matching, EGM enables training of arbitrary continuous-time Markov processes, e.g., diffusion, flow, and jump, and can generate data from continuous, discrete, and a mixture of two modalities. To this end, we propose estimating the generator matching loss using self-normalized importance sampling with an additional bootstrapping trick to reduce variance in the importance weight. We validate EGM on both discrete and multimodal tasks up to 100 and 20 dimensions, respectively.
Authors: J\"org K. H. Franke, Urs Spiegelhalter, Marianna Nezhurina, Jenia Jitsev, Frank Hutter, Michael Hefenbrock
Abstract: The successful training of deep neural networks requires addressing challenges such as overfitting, numerical instabilities leading to divergence, and increasing variance in the residual stream. A common solution is to apply regularization and normalization techniques that usually require tuning additional hyperparameters. An alternative is to force all parameters and representations to lie on a hypersphere. This removes the need for regularization and increases convergence speed, but comes with additional costs. In this work, we propose a more holistic, approximate normalization via simple scalar multiplications motivated by the tight concentration of the norms of high-dimensional random vectors. Additionally, instead of applying strict normalization for the parameters, we constrain their norms. These modifications remove the need for weight decay and learning rate warm-up as well, but do not increase the total number of normalization layers. Our experiments with transformer architectures show up to 40% faster convergence compared to GPT models with QK normalization, with only 3% additional runtime cost. When deriving scaling laws, we found that our method enables training with larger batch sizes while preserving the favorable scaling characteristics of classic GPT architectures.
Authors: Tyler Chen, Akshay Seshadri, Mattia J. Villani, Pradeep Niroula, Shouvanik Chakrabarti, Archan Ray, Pranav Deshpande, Romina Yalovetzky, Marco Pistoia, Niraj Kumar
Abstract: Shapley values have emerged as a critical tool for explaining which features impact the decisions made by machine learning models. However, computing exact Shapley values is difficult, generally requiring an exponential (in the feature dimension) number of model evaluations. To address this, many model-agnostic randomized estimators have been developed, the most influential and widely used being the KernelSHAP method (Lundberg & Lee, 2017). While related estimators such as unbiased KernelSHAP (Covert & Lee, 2021) and LeverageSHAP (Musco & Witter, 2025) are known to satisfy theoretical guarantees, bounds for KernelSHAP have remained elusive. We describe a broad and unified framework that encompasses KernelSHAP and related estimators constructed using both with and without replacement sampling strategies. We then prove strong non-asymptotic theoretical guarantees that apply to all estimators from our framework. This provides, to the best of our knowledge, the first theoretical guarantees for KernelSHAP and sheds further light on tradeoffs between existing estimators. Through comprehensive benchmarking on small and medium dimensional datasets for Decision-Tree models, we validate our approach against exact Shapley values, consistently achieving low mean squared error with modest sample sizes. Furthermore, we make specific implementation improvements to enable scalability of our methods to high-dimensional datasets. Our methods, tested on datasets such MNIST and CIFAR10, provide consistently better results compared to the KernelSHAP library.
Authors: Konstantin Riedl, Justin Sirignano, Konstantinos Spiliopoulos
Abstract: Many engineering and scientific fields have recently become interested in modeling terms in partial differential equations (PDEs) with neural networks, which requires solving the inverse problem of learning neural network terms from observed data in order to approximate missing or unresolved physics in the PDE model. The resulting neural-network PDE model, being a function of the neural network parameters, can be calibrated to the available ground truth data by optimizing over the PDE using gradient descent, where the gradient is evaluated in a computationally efficient manner by solving an adjoint PDE. These neural PDE models have emerged as an important research area in scientific machine learning. In this paper, we study the convergence of the adjoint gradient descent optimization method for training neural PDE models in the limit where both the number of hidden units and the training time tend to infinity. Specifically, for a general class of nonlinear parabolic PDEs with a neural network embedded in the source term, we prove convergence of the trained neural-network PDE solution to the target data (i.e., a global minimizer). The global convergence proof poses a unique mathematical challenge that is not encountered in finite-dimensional neural network convergence analyses due to (i) the neural network training dynamics involving a non-local neural network kernel operator in the infinite-width hidden layer limit where the kernel lacks a spectral gap for its eigenvalues and (ii) the nonlinearity of the limit PDE system, which leads to a non-convex optimization problem in the neural network function even in the infinite-width hidden layer limit (unlike in typical neural network training cases where the optimization problem becomes convex in the large neuron limit). The theoretical results are illustrated and empirically validated by numerical studies.
Authors: Anirudh Varanasi, Robin Degraeve, Philippe Roussel, Clement Merckling
Abstract: Random telegraph noise is a prevalent variability phenomenon in nanoelectronic devices, arising from stochastic carrier exchange at defect sites and critically impacting device reliability and performance. Conventional analysis techniques often rely on restrictive assumptions or manual interventions, limiting their applicability to complex, noisy datasets. Here, we introduce RTNinja, a generalized, fully automated machine learning framework for the unsupervised analysis of random telegraph noise signals. RTNinja deconvolves complex signals to identify the number and characteristics of hidden individual sources without requiring prior knowledge of the system. The framework comprises two modular components: LevelsExtractor, which uses Bayesian inference and model selection to denoise and discretize the signal, and SourcesMapper, which infers source configurations through probabilistic clustering and optimization. To evaluate performance, we developed a Monte Carlo simulator that generates labeled datasets spanning broad signal-to-noise ratios and source complexities; across 7000 such datasets, RTNinja consistently demonstrated high-fidelity signal reconstruction and accurate extraction of source amplitudes and activity patterns. Our results demonstrate that RTNinja offers a robust, scalable, and device-agnostic tool for random telegraph noise characterization, enabling large-scale statistical benchmarking, reliability-centric technology qualification, predictive failure modeling, and device physics exploration in next-generation nanoelectronics.
Authors: Scott Pesme, Giacomo Meanti, Michael Arbel, Julien Mairal
Abstract: Denoiser models have become powerful tools for inverse problems, enabling the use of pretrained networks to approximate the score of a smoothed prior distribution. These models are often used in heuristic iterative schemes aimed at solving Maximum a Posteriori (MAP) optimisation problems, where the proximal operator of the negative log-prior plays a central role. In practice, this operator is intractable, and practitioners plug in a pretrained denoiser as a surrogate-despite the lack of general theoretical justification for this substitution. In this work, we show that a simple algorithm, closely related to several used in practice, provably converges to the proximal operator under a log-concavity assumption on the prior $p$. We show that this algorithm can be interpreted as a gradient descent on smoothed proximal objectives. Our analysis thus provides a theoretical foundation for a class of empirically successful but previously heuristic methods.
Authors: Xuanhao Mu, G\"okhan Demirel, Yuzhe Zhang, Jianlei Liu, Thorsten Schlachter, Veit Hagenmeyer
Abstract: To bridge the temporal granularity gap in energy network design and operation based on Energy System Models, resampling of time series is required. While conventional upsampling methods are computationally efficient, they often result in significant information loss or increased noise. Advanced models such as time series generation models, Super-Resolution models and imputation models show potential, but also face fundamental challenges. The goal of time series generative models is to learn the distribution of the original data to generate high-resolution series with similar statistical characteristics. This is not entirely consistent with the definition of upsampling. Time series Super-Resolution models or imputation models can degrade the accuracy of upsampling because the input low-resolution time series are sparse and may have insufficient context. Moreover, such models usually rely on supervised learning paradigms. This presents a fundamental application paradox: their training requires the high-resolution time series that is intrinsically absent in upsampling application scenarios. To address the mentioned upsampling issue, this paper introduces a new method utilizing Generative Adversarial Transformers (GATs), which can be trained without access to any ground-truth high-resolution data. Compared with conventional interpolation methods, the introduced method can reduce the root mean square error (RMSE) of upsampling tasks by 10%, and the accuracy of a model predictive control (MPC) application scenario is improved by 13%.
Authors: Zhendong Yang, Jie Wang, Liansong Zong, Xiaorong Liu, Quan Qian, Shiqian Chen
Abstract: Few-Shot Class-Incremental Fault Diagnosis (FSC-FD), which aims to continuously learn from new fault classes with only a few samples without forgetting old ones, is critical for real-world industrial systems. However, this challenging task severely amplifies the issues of catastrophic forgetting of old knowledge and overfitting on scarce new data. To address these challenges, this paper proposes a novel framework built upon Dual-Granularity Representations, termed the Dual-Granularity Guidance Network (DGGN). Our DGGN explicitly decouples feature learning into two parallel streams: 1) a fine-grained representation stream, which utilizes a novel Multi-Order Interaction Aggregation module to capture discriminative, class-specific features from the limited new samples. 2) a coarse-grained representation stream, designed to model and preserve general, class-agnostic knowledge shared across all fault types. These two representations are dynamically fused by a multi-semantic cross-attention mechanism, where the stable coarse-grained knowledge guides the learning of fine-grained features, preventing overfitting and alleviating feature conflicts. To further mitigate catastrophic forgetting, we design a Boundary-Aware Exemplar Prioritization strategy. Moreover, a decoupled Balanced Random Forest classifier is employed to counter the decision boundary bias caused by data imbalance. Extensive experiments on the TEP benchmark and a real-world MFF dataset demonstrate that our proposed DGGN achieves superior diagnostic performance and stability compared to state-of-the-art FSC-FD approaches. Our code is publicly available at https://github.com/MentaY/DGGN
Authors: Manish Nagaraj, Deepak Ravikumar, Kaushik Roy
Abstract: Deep learning models achieve state-of-the-art performance across domains but face scalability challenges in real-time or resource-constrained scenarios. To address this, we propose Correlation of Loss Differences (CLD), a simple and scalable metric for coreset selection that identifies the most impactful training samples by measuring their alignment with the loss trajectories of a held-out validation set. CLD is highly efficient, requiring only per-sample loss values computed at training checkpoints, and avoiding the costly gradient and curvature computations used in many existing subset selection methods. We develop a general theoretical framework that establishes convergence guarantees for CLD-based coresets, demonstrating that the convergence error is upper-bounded by the alignment of the selected samples and the representativeness of the validation set. On CIFAR-100 and ImageNet-1k, CLD-based coresets typically outperform or closely match state-of-the-art methods across subset sizes, and remain within 1% of more computationally expensive baselines even when not leading. CLD transfers effectively across architectures (ResNet, VGG, DenseNet), enabling proxy-to-target selection with <1% degradation. Moreover, CLD is stable when using only early checkpoints, incurring negligible accuracy loss. Finally, CLD exhibits inherent bias reduction via per-class validation alignment, obviating the need for additional stratified sampling. Together, these properties make CLD a principled, efficient, stable, and transferable tool for scalable dataset optimization.
Authors: Weicao Deng, Sangwoo Park, Min Li, Osvaldo Simeone
Abstract: Reliable uncertainty quantification is critical for trustworthy AI. Conformal Prediction (CP) provides prediction sets with distribution-free coverage guarantees, but its two main variants face complementary limitations. Split CP (SCP) suffers from data inefficiency due to dataset partitioning, while full CP (FCP) improves data efficiency at the cost of prohibitive retraining complexity. Recent approaches based on meta-learning or in-context learning (ICL) partially mitigate these drawbacks. However, they rely on training procedures not specifically tailored to CP, which may yield large prediction sets. We introduce an efficient FCP framework, termed enhanced ICL-based FCP (E-ICL+FCP), which employs a permutation-invariant Transformer-based ICL model trained with a CP-aware loss. By simulating the multiple retrained models required by FCP without actual retraining, E-ICL+FCP preserves coverage while markedly reducing both inefficiency and computational overhead. Experiments on synthetic and real tasks demonstrate that E-ICL+FCP attains superior efficiency-coverage trade-offs compared to existing SCP and FCP baselines.
Authors: Ibne Farabi Shihab, Sanjeda Akter, Anuj Sharma
Abstract: We introduce the first differentiable approximation of range-partition entropy, a complexity measure from computational geometry that directly bounds algorithmic runtime. Unlike architectural modifications, our method is a complementary regularizer that provides orthogonal efficiency gains when combined with existing optimizations. We establish theoretical guarantees in computational geometry, achieving 4--5$\times$ provable speedups on convex hull and triangulation with $<$0.2\% error. On ImageNet-1K with ViT-Base, entropy regularization achieves 80.1\% top-1 accuracy at 80\% sparsity (1.60$\times$ standalone speedup), and when combined with FlashAttention yields 2.07$\times$ speedup versus 1.63$\times$ for FlashAttention alone. On large language models (LLaMA-2 7B, Mistral-7B, Phi-2), we achieve 1.48--1.60$\times$ inference speedups at 70--75\% sparsity with minimal quality degradation (ROUGE-L drops of 0.3--0.4 points, perplexity increase of 0.9). Unlike prior regularization methods that target output distributions, we directly minimize representation complexity, yielding both efficiency gains and improved robustness through semantically structured sparsity patterns (IoU 0.73 vs 0.41 for magnitude pruning, CIFAR-100-C mCE 48.7 vs 55.4). Benefits are strongest for geometry and vision transformers, with more modest but measurable gains on LLMs, demonstrating that complexity regularization offers a principled pathway to joint efficiency-robustness optimization.
Authors: Sanjeda Akter, Ibne Farabi Shihab, Anuj Sharma
Abstract: Large language models often produce confident but incorrect outputs, creating a critical need for reliable uncertainty quantification with formal abstention guarantees. We introduce information-lift certificates that compare model probabilities to a skeleton baseline, accumulating evidence through sub-gamma PAC-Bayes bounds that remain valid under heavy-tailed distributions where standard concentration inequalities fail. On eight diverse datasets, our method achieves 77.0\% coverage at 2\% risk, outperforming recent baselines by 10.0 percentage points on average. In high-stakes scenarios, we block 96\% of critical errors compared to 18-31\% for entropy-based methods. While our frequency-based certification does not guarantee severity-weighted safety and depends on skeleton quality, performance degrades gracefully under distributional shifts, making the approach practical for real-world deployment.
Authors: Bishnu Bhusal, Manoj Acharya, Ramneet Kaur, Colin Samplawski, Anirban Roy, Adam D. Cobb, Rohit Chadha, Susmit Jha
Abstract: Large language models (LLMs) have significantly transformed natural language understanding and generation, but they raise privacy concerns due to potential exposure of sensitive information. Studies have highlighted the risk of information leakage, where adversaries can extract sensitive information embedded in the prompts. In this work, we introduce a novel private prediction framework for generating high-quality synthetic text with strong privacy guarantees. Our approach leverages the Differential Privacy (DP) framework to ensure worst-case theoretical bounds on information leakage without requiring any fine-tuning of the underlying models. The proposed method performs inference on private records and aggregates the resulting per-token output distributions. This enables the generation of longer and coherent synthetic text while maintaining privacy guarantees. Additionally, we propose a simple blending operation that combines private and public inference to further enhance utility. Empirical evaluations demonstrate that our approach outperforms previous state-of-the-art methods on in-context-learning (ICL) tasks, making it a promising direction for privacy-preserving text generation while maintaining high utility. Our code is available at https://github.com/bhusalb/privacy-preserving-icl.
Authors: Longxuan Yu, Xing Shi, Xianghao Kong, Tong Jia, Greg Ver Steeg
Abstract: Mutual information (MI) is one of the most general ways to measure relationships between random variables, but estimating this quantity for complex systems is challenging. Denoising diffusion models have recently set a new bar for density estimation, so it is natural to consider whether these methods could also be used to improve MI estimation. Using the recently introduced information-theoretic formulation of denoising diffusion models, we show the diffusion models can be used in a straightforward way to estimate MI. In particular, the MI corresponds to half the gap in the Minimum Mean Square Error (MMSE) between conditional and unconditional diffusion, integrated over all Signal-to-Noise-Ratios (SNRs) in the noising process. Our approach not only passes self-consistency tests but also outperforms traditional and score-based diffusion MI estimators. Furthermore, our method leverages adaptive importance sampling to achieve scalable MI estimation, while maintaining strong performance even when the MI is high.
Authors: Zhichao Sheng, Shilin Zhou, Chen Gong, Zhenghua Li
Abstract: Large Audio Language Models (LALMs), powered by the chain-of-thought (CoT) paradigm, have shown remarkable reasoning capabilities. Intuitively, different problems often require varying depths of reasoning. While some methods can determine whether to reason for a given problem, they typically lack a fine-grained mechanism to modulate how much to reason. This often results in a ``one-size-fits-all'' reasoning depth, which generates redundant overthinking for simple questions while failing to allocate sufficient thought to complex ones. In this paper, we conduct an in-depth analysis of LALMs and find that an effective and efficient LALM should reason smartly by adapting its reasoning depth to the problem's complexity. To achieve this, we propose a difficulty-adaptive reasoning method for LALMs. Specifically, we propose a reward function that dynamically links reasoning length to the model's perceived problem difficulty. This reward encourages shorter, concise reasoning for easy tasks and more elaborate, in-depth reasoning for complex ones. Extensive experiments demonstrate that our method is both effective and efficient, simultaneously improving task performance and significantly reducing the average reasoning length. Further analysis on reasoning structure paradigm offers valuable insights for future work.
Authors: Sameep Chattopadhyay, Nikhil Karamchandani, Sharayu Moharir
Abstract: Online learning to rank (OLTR) plays a critical role in information retrieval and machine learning systems, with a wide range of applications in search engines and content recommenders. However, despite their extensive adoption, the susceptibility of OLTR algorithms to coordinated adversarial attacks remains poorly understood. In this work, we present a novel framework for attacking some of the widely used OLTR algorithms. Our framework is designed to promote a set of target items so that they appear in the list of top-K recommendations for T - o(T) rounds, while simultaneously inducing linear regret in the learning algorithm. We propose two novel attack strategies: CascadeOFA for CascadeUCB1 and PBMOFA for PBM-UCB . We provide theoretical guarantees showing that both strategies require only O(log T) manipulations to succeed. Additionally, we supplement our theoretical analysis with empirical results on real-world data.
Authors: Jintao Zhang, Haoxu Wang, Kai Jiang, Shuo Yang, Kaiwen Zheng, Haocheng Xi, Ziteng Wang, Hongzhou Zhu, Min Zhao, Ion Stoica, Joseph E. Gonzalez, Jun Zhu, Jianfei Chen
Abstract: In Diffusion Transformer (DiT) models, particularly for video generation, attention latency is a major bottleneck due to the long sequence length and the quadratic complexity. We find that attention weights can be separated into two parts: a small fraction of large weights with high rank and the remaining weights with very low rank. This naturally suggests applying sparse acceleration to the first part and low-rank acceleration to the second. Based on this finding, we propose SLA (Sparse-Linear Attention), a trainable attention method that fuses sparse and linear attention to accelerate diffusion models. SLA classifies attention weights into critical, marginal, and negligible categories, applying O(N^2) attention to critical weights, O(N) attention to marginal weights, and skipping negligible ones. SLA combines these computations into a single GPU kernel and supports both forward and backward passes. With only a few fine-tuning steps using SLA, DiT models achieve a 20x reduction in attention computation, resulting in significant acceleration without loss of generation quality. Experiments show that SLA reduces attention computation by 95% without degrading end-to-end generation quality, outperforming baseline methods. In addition, we implement an efficient GPU kernel for SLA, which yields a 13.7x speedup in attention computation and a 2.2x end-to-end speedup in video generation on Wan2.1-1.3B. The code is available at https://github.com/thu-ml/SLA.
Authors: Daniel Jason Tan, Jiayang Chen, Dilruk Perera, Kay Choong See, Mengling Feng
Abstract: ICU enteral feeding remains sub-optimal due to limited personalization and uncertainty about appropriate calorie, protein, and fluid targets, particularly under rapidly changing metabolic demands and heterogeneous patient responses. This study introduces DeepEN, a reinforcement learning (RL)-based framework that personalizes enteral nutrition (EN) dosing for critically ill patients using electronic health record data. DeepEN was trained on over 11,000 ICU patients from the MIMIC-IV database to generate 4-hourly, patient-specific targets for caloric, protein, and fluid intake. The model's state space integrates demographics, comorbidities, vital signs, laboratory results, and prior interventions relevant to nutritional management, while its reward function balances short-term physiological and nutrition-related goals with long-term survival. A dueling double deep Q-network with Conservative Q-Learning regularization is used to ensure safe and reliable policy learning from retrospective data. DeepEN achieved a 3.7 $\pm$ 0.17 percentage-point absolute reduction in estimated mortality compared with the clinician policy (18.8% vs 22.5%) and higher expected returns compared with guideline-based dosing (11.89 vs 8.11), with improvements in key nutritional biomarkers. U-shaped associations between deviations from clinician dosing and mortality suggest that the learned policy aligns with high-value clinician actions while diverging from suboptimal ones. These findings demonstrate the feasibility of conservative offline RL for individualized EN therapy and suggest that data-driven personalization may improve outcomes beyond guideline- or heuristic-based approaches.
Authors: Kun Chen, Peng Shi, Haibo Qiu, Zhixiong Zeng, Siqi Yang, Wenji Mao, Lin Ma
Abstract: Reinforcement learning (RL) with verifiable rewards has recently catalyzed a wave of "MLLM-r1" approaches that bring RL to vision language models. Most representative paradigms begin with a cold start, typically employing supervised fine-tuning (SFT), to initialize the policy before RL. However, SFT-based cold start adopts the reasoning paradigm intertwined with task solution and output format, which may induce instruction-style overfitting, weakens out-of-distribution generalization, and ultimately affects downstream RL. We revisit the cold start along two views, its training method and data construction, and introduce the Generalization Factor (GF) coefficient to quantify the generalization capability under different methods. Our empirical study finds that preference-based training methods (e.g. DPO) generalizes better than SFT-based methods in cold start. Motivated by this, we propose SPECS-a Self-distilled, Preference-based Cold Start framework that decouples multimodal learning: (1) generates introspective preference data pairs via self-distillation, avoiding reliance on larger teachers or manual annotation; (2) performs preference-based training to learn, focusing on shallow, transferable surface-form criteria (format, structure, style) rather than memorizing content; and (3) hands off to RL with verifiable rewards for deep reasoning results. Experimental results across multiple multimodal benchmarks show that our decoupling learning framework yields consistent performance gains over strong baselines, improving MEGA-Bench by 4.1% and MathVista by 12.2%. Additional experiments indicate that SPECS contributes to reducing in-distribution "stuckness," improving exploration, stabilizing training, and raising the performance ceiling.
Authors: Davide Marincione, Donato Crisostomi, Roberto Dessi, Emanuele Rodol\`a, Emanuele Rossi
Abstract: Foundation models capable of generalizing across species and tasks represent a promising new frontier in bioacoustics, with NatureLM being one of the most prominent examples. While its domain-specific fine-tuning yields strong performance on bioacoustic benchmarks, we observe that it also introduces trade-offs in instruction-following flexibility. For instance, NatureLM achieves high accuracy when prompted for either the common or scientific name individually, but its accuracy drops significantly when both are requested in a single prompt. We address this by applying a simple model merging strategy that interpolates NatureLM with its base language model, recovering instruction-following capabilities with minimal loss of domain expertise. Finally, we show that the merged model exhibits markedly stronger zero-shot generalization, achieving over a 200% relative improvement and setting a new state-of-the-art in closed-set zero-shot classification of unseen species.
Authors: Richard Goldman, Varun Komperla, Thomas Ploetz, Harish Haresamudram
Abstract: A promising alternative to the computationally expensive Neural Architecture Search (NAS) involves the development of Zero Cost Proxies (ZCPs), which correlate well with trained performance, but can be computed through a single forward/backward pass on a randomly sampled batch of data. In this paper, we investigate the effectiveness of ZCPs for HAR on six benchmark datasets, and demonstrate that they discover network architectures that obtain within 5% of performance attained by full-scale training involving 1500 randomly sampled architectures. This results in substantial computational savings as high-performing architectures can be discovered with minimal training. Our experiments not only introduce ZCPs to sensor-based HAR, but also demonstrate that they are robust to data noise, further showcasing their suitability for practical scenarios.
Authors: Paul Strang, Zacharie Al\`es, C\^ome Bissuel, Safia Kedad-Sidhoum, Emmanuel Rachelson
Abstract: Mixed-Integer Linear Programming (MILP) lies at the core of many real-world combinatorial optimization (CO) problems, traditionally solved by branch-and-bound (B&B). A key driver influencing B&B solvers efficiency is the variable selection heuristic that guides branching decisions. Looking to move beyond static, hand-crafted heuristics, recent work has explored adapting traditional reinforcement learning (RL) algorithms to the B&B setting, aiming to learn branching strategies tailored to specific MILP distributions. In parallel, RL agents have achieved remarkable success in board games, a very specific type of combinatorial problems, by leveraging environment simulators to plan via Monte Carlo Tree Search (MCTS). Building on these developments, we introduce Plan-and-Branch-and-Bound (PlanB&B), a model-based reinforcement learning (MBRL) agent that leverages a learned internal model of the B&B dynamics to discover improved branching strategies. Computational experiments empirically validate our approach, with our MBRL branching agent outperforming previous state-of-the-art RL methods across four standard MILP benchmarks.
Authors: Chenyue Liu, Ali Mostafavi
Abstract: Current wildfire risk assessments rely on coarse hazard maps and opaque machine learning models that optimize regional accuracy while sacrificing interpretability at the decision scale. WildfireGenome addresses these gaps through three components: (1) fusion of seven federal wildfire indicators into a sign-aligned, PCA-based composite risk label at H3 Level-8 resolution; (2) Random Forest classification of local wildfire risk; and (3) SHAP and ICE/PDP analyses to expose county-specific nonlinear driver relationships. Across seven ecologically diverse U.S. counties, models achieve accuracies of 0.755-0.878 and Quadratic Weighted Kappa up to 0.951, with principal components explaining 87-94% of indicator variance. Transfer tests show reliable performance between ecologically similar regions but collapse across dissimilar contexts. Explanations consistently highlight needleleaf forest cover and elevation as dominant drivers, with risk rising sharply at 30-40% needleleaf coverage. WildfireGenome advances wildfire risk assessment from regional prediction to interpretable, decision-scale analytics that guide vegetation management, zoning, and infrastructure planning.
Authors: Qing Yao, Lijian Gao, Qirong Mao, Ming Dong
Abstract: Diffusion models serve as a powerful generative framework for solving inverse problems. However, they still face two key challenges: 1) the distortion-perception tradeoff, where improving perceptual quality often degrades reconstruction fidelity, and 2) the exposure bias problem, where the training-inference input mismatch leads to prediction error accumulation and reduced reconstruction quality. In this work, we propose the Regularized Schr\"odinger Bridge (RSB), an adaptation of Schr\"odinger Bridge tailored for inverse problems that addresses the above limitations. RSB employs a novel regularized training strategy that perturbs both the input states and targets, effectively mitigating exposure bias by exposing the model to simulated prediction errors and also alleviating distortion by well-designed interpolation via the posterior mean. Extensive experiments on two typical inverse problems for speech enhancement demonstrate that RSB outperforms state-of-the-art methods, significantly improving distortion metrics and effectively reducing exposure bias.
Authors: Sebasti\'an Andr\'es Cajas Ord\'o\~nez, Luis Fernando Torres Torres, Mackenzie J. Meni, Carlos Andr\'es Duran Paredes, Eric Arazo, Cristian Bosch, Ricardo Simon Carbajo, Yuan Lai, Leo Anthony Celi
Abstract: Deploying deep neural networks on resource-constrained devices faces two critical challenges: maintaining accuracy under aggressive quantization while ensuring predictable inference latency. We present a curiosity-driven quantized Mixture-of-Experts framework that addresses both through Bayesian epistemic uncertainty-based routing across heterogeneous experts (BitNet ternary, 1-16 bit BitLinear, post-training quantization). Evaluated on audio classification benchmarks (ESC-50, Quinn, UrbanSound8K), our 4-bit quantization maintains 99.9 percent of 16-bit accuracy (0.858 vs 0.859 F1) with 4x compression and 41 percent energy savings versus 8-bit. Crucially, curiosity-driven routing reduces MoE latency variance by 82 percent (p = 0.008, Levene's test) from 230 ms to 29 ms standard deviation, enabling stable inference for battery-constrained devices. Statistical analysis confirms 4-bit/8-bit achieve practical equivalence with full precision (p > 0.05), while MoE architectures introduce 11 percent latency overhead (p < 0.001) without accuracy gains. At scale, deployment emissions dominate training by 10000x for models serving more than 1,000 inferences, making inference efficiency critical. Our information-theoretic routing demonstrates that adaptive quantization yields accurate (0.858 F1, 1.2M params), energy-efficient (3.87 F1/mJ), and predictable edge models, with simple 4-bit quantized architectures outperforming complex MoE for most deployments.
Authors: Byoungjun Park, Pedro Porto Buarque de Gusm\~ao, Dongjin Ji, Minhoe Kim
Abstract: Federated learning is a promising paradigm that utilizes distributed client resources while preserving data privacy. Most existing FL approaches assume clients possess labeled data, however, in real-world scenarios, client-side labels are often unavailable. Semi-supervised Federated learning, where only the server holds labeled data, addresses this issue. However, it experiences significant performance degradation as the number of labeled data decreases. To tackle this problem, we propose \textit{CATCHFed}, which introduces client-aware adaptive thresholds considering class difficulty, hybrid thresholds to enhance pseudo-label quality, and utilizes unpseudo-labeled data for consistency regularization. Extensive experiments across various datasets and configurations demonstrate that CATCHFed effectively leverages unlabeled client data, achieving superior performance even in extremely limited-label settings.
Authors: Zhengxin Zhang, Chengyu Huang, Aochong Oliver Li, Claire Cardie
Abstract: Large Language Models (LLMs) have achieved remarkable progress through Reinforcement Learning with Verifiable Rewards (RLVR), yet still rely heavily on external supervision (e.g., curated labels). Adversarial learning, particularly through self-play, offers a promising alternative that enables models to iteratively learn from themselves - thus reducing reliance on external supervision. Dual-play extends adversarial learning by assigning specialized roles to two models and training them against each other, fostering sustained competition and mutual evolution. Despite its promise, adapting dual-play training to LLMs remains limited, largely due to their susceptibility to reward hacking and training instability. In this paper, we introduce PasoDoble, a novel LLM dual-play framework. PasoDoble adversarially trains two models initialized from the same base model: a Proposer, which generates challenging questions with ground-truth answers, and a Solver, which attempts to solve them. We enrich the Proposer with knowledge from a pre-training dataset to ensure the questions' quality and diversity. To avoid reward hacking, the Proposer is rewarded for producing only valid questions that push the Solver's limit, while the Solver is rewarded for solving them correctly, and both are updated jointly. To further enhance training stability, we introduce an optional offline paradigm that decouples Proposer and Solver updates, alternately updating each for several steps while holding the other fixed. Notably, PasoDoble operates without supervision during training. Experimental results show that PasoDoble can improve the reasoning performance of LLMs. Our project page is available at https://hcy123902.github.io/PasoDoble.
Authors: Wanlong Fang, Tianle Zhang, Alvin Chan
Abstract: Multimodal learning often relies on aligning representations across modalities to enable effective information integration, an approach traditionally assumed to be universally beneficial. However, prior research has primarily taken an observational approach, examining naturally occurring alignment in multimodal data and exploring its correlation with model performance, without systematically studying the direct effects of explicitly enforced alignment between representations of different modalities. In this work, we investigate how explicit alignment influences both model performance and representation alignment under different modality-specific information structures. Specifically, we introduce a controllable contrastive learning module that enables precise manipulation of alignment strength during training, allowing us to explore when explicit alignment improves or hinders performance. Our results on synthetic and real datasets under different data characteristics show that the impact of explicit alignment on the performance of unimodal models is related to the characteristics of the data: the optimal level of alignment depends on the amount of redundancy between the different modalities. We identify an optimal alignment strength that balances modality-specific signals and shared redundancy in the mixed information distributions. This work provides practical guidance on when and how explicit alignment should be applied to achieve optimal unimodal encoder performance.
Authors: Meng Zhu, Quan Xiao, Weidong Min
Abstract: Since the 21st century, artificial intelligence has been leading a new round of industrial revolution. Under the training framework, the optimization algorithm aims to stably converge high-dimensional optimization to local and even global minima. Entering the era of large language models, although the scale of model parameters and data has increased, Adam remains the mainstream optimization algorithm. However, compared with stochastic gradient descent (SGD) based optimization algorithms, Adam is more likely to converge to non-flat minima. To address this issue, the AdamX algorithm is proposed. Its core innovation lies in the proposition of a novel type of second-order moment estimation exponential decay rate, which gradually weakens the learning step correction strength as training progresses, and degrades to SGD in the stable training period, thereby improving the stability of training in the stable period and possibly enhancing generalization ability. Experimental results show that our second-order moment estimation exponential decay rate is better than the current second-order moment estimation exponential decay rate, and AdamX can stably outperform Adam and its variants in terms of performance. Our code is open-sourced at https://github.com/mengzhu0308/AdamX.
Authors: Federico Battini
Abstract: Accurate renewable energy forecasting is essential to reduce dependence on fossil fuels and enabling grid decarbonization. However, current approaches fail to effectively integrate the rich spatial context of weather patterns with their temporal evolution. This work introduces a novel approach that treats weather maps as tokens in transformer sequences to predict renewable energy. Hourly weather maps are encoded as spatial tokens using a lightweight convolutional neural network, and then processed by a transformer to capture temporal dynamics across a 45-hour forecast horizon. Despite disadvantages in input initialization, evaluation against ENTSO-E operational forecasts shows a reduction in RMSE of about 60% and 20% for wind and solar respectively. A live dashboard showing daily forecasts is available at: https://www.sardiniaforecast.ifabfoundation.it.
Authors: Hao Qian, Shikui Tu, Lei Xu
Abstract: Diffusion and flow matching models have recently emerged as promising approaches for peptide binder design. Despite their progress, these models still face two major challenges. First, categorical sampling of discrete residue types collapses their continuous parameters into onehot assignments, while continuous variables (e.g., atom positions) evolve smoothly throughout the generation process. This mismatch disrupts the update dynamics and results in suboptimal performance. Second, current models assume unimodal distributions for side-chain torsion angles, which conflicts with the inherently multimodal nature of side chain rotameric states and limits prediction accuracy. To address these limitations, we introduce PepBFN, the first Bayesian flow network for full atom peptide design that directly models parameter distributions in fully continuous space. Specifically, PepBFN models discrete residue types by learning their continuous parameter distributions, enabling joint and smooth Bayesian updates with other continuous structural parameters. It further employs a novel Gaussian mixture based Bayesian flow to capture the multimodal side chain rotameric states and a Matrix Fisher based Riemannian flow to directly model residue orientations on the $\mathrm{SO}(3)$ manifold. Together, these parameter distributions are progressively refined via Bayesian updates, yielding smooth and coherent peptide generation. Experiments on side chain packing, reverse folding, and binder design tasks demonstrate the strong potential of PepBFN in computational peptide design.
Authors: Abolfazl Younesi, Leon Kiss, Zahra Najafabadi Samani, Juan Aznar Poveda, Thomas Fahringer
Abstract: Federated learning (FL) enables collaborative model training while preserving data privacy. However, it remains vulnerable to malicious clients who compromise model integrity through Byzantine attacks, data poisoning, or adaptive adversarial behaviors. Existing defense mechanisms rely on static thresholds and binary classification, failing to adapt to evolving client behaviors in real-world deployments. We propose FLARE, an adaptive reputation-based framework that transforms client reliability assessment from binary decisions to a continuous, multi-dimensional trust evaluation. FLARE integrates: (i) a multi-dimensional reputation score capturing performance consistency, statistical anomaly indicators, and temporal behavior, (ii) a self-calibrating adaptive threshold mechanism that adjusts security strictness based on model convergence and recent attack intensity, (iii) reputation-weighted aggregation with soft exclusion to proportionally limit suspicious contributions rather than eliminating clients outright, and (iv) a Local Differential Privacy (LDP) mechanism enabling reputation scoring on privatized client updates. We further introduce a highly evasive Statistical Mimicry (SM) attack, a benchmark adversary that blends honest gradients with synthetic perturbations and persistent drift to remain undetected by traditional filters. Extensive experiments with 100 clients on MNIST, CIFAR-10, and SVHN demonstrate that FLARE maintains high model accuracy and converges faster than state-of-the-art Byzantine-robust methods under diverse attack types, including label flipping, gradient scaling, adaptive attacks, ALIE, and SM. FLARE improves robustness by up to 16% and preserves model convergence within 30% of the non-attacked baseline, while achieving strong malicious-client detection performance with minimal computational overhead. https://github.com/Anonymous0-0paper/FLARE
Authors: Physical Intelligence, Ali Amin, Raichelle Aniceto, Ashwin Balakrishna, Kevin Black, Ken Conley, Grace Connors, James Darpinian, Karan Dhabalia, Jared DiCarlo, Danny Driess, Michael Equi, Adnan Esmail, Yunhao Fang, Chelsea Finn, Catherine Glossop, Thomas Godden, Ivan Goryachev, Lachy Groom, Hunter Hancock, Karol Hausman, Gashon Hussein, Brian Ichter, Szymon Jakubczak, Rowan Jen, Tim Jones, Ben Katz, Liyiming Ke, Chandra Kuchi, Marinda Lamb, Devin LeBlanc, Sergey Levine, Adrian Li-Bell, Yao Lu, Vishnu Mano, Mohith Mothukuri, Suraj Nair, Karl Pertsch, Allen Z. Ren, Charvi Sharma, Lucy Xiaoyang Shi, Laura Smith, Jost Tobias Springenberg, Kyle Stachowicz, Will Stoeckle, Alex Swerdlow, James Tanner, Marcel Torne, Quan Vuong, Anna Walling, Haohuan Wang, Blake Williams, Sukwon Yoo, Lili Yu, Ury Zhilinsky, Zhiyuan Zhou
Abstract: We study how vision-language-action (VLA) models can improve through real-world deployments via reinforcement learning (RL). We present a general-purpose method, RL with Experience and Corrections via Advantage-conditioned Policies (RECAP), that provides for RL training of VLAs via advantage conditioning. Our method incorporates heterogeneous data into the self-improvement process, including demonstrations, data from on-policy collection, and expert teleoperated interventions provided during autonomous execution. RECAP starts by pre-training a generalist VLA with offline RL, which we call $\pi^{*}_{0.6}$, that can then be specialized to attain high performance on downstream tasks through on-robot data collection. We show that the $\pi^{*}_{0.6}$ model trained with the full RECAP method can fold laundry in real homes, reliably assemble boxes, and make espresso drinks using a professional espresso machine. On some of the hardest tasks, RECAP more than doubles task throughput and roughly halves the task failure rate.
Authors: Hari Dahal, Wei Liu, Yangyang Xu
Abstract: We give a damped proximal augmented Lagrangian method (DPALM) for solving problems with a weakly-convex objective and convex linear/nonlinear constraints. Instead of taking a full stepsize, DPALM adopts a damped dual stepsize to ensure the boundedness of dual iterates. We show that DPALM can produce a (near) $\vareps$-KKT point within $O(\vareps^{-2})$ outer iterations if each DPALM subproblem is solved to a proper accuracy. In addition, we establish overall iteration complexity of DPALM when the objective is either a regularized smooth function or in a regularized compositional form. For the former case, DPALM achieves the complexity of $\widetilde{\mathcal{O}}\left(\varepsilon^{-2.5} \right)$ to produce an $\varepsilon$-KKT point by applying an accelerated proximal gradient (APG) method to each DPALM subproblem. For the latter case, the complexity of DPALM is $\widetilde{\mathcal{O}}\left(\varepsilon^{-3} \right)$ to produce a near $\varepsilon$-KKT point by using an APG to solve a Moreau-envelope smoothed version of each subproblem. Our outer iteration complexity and the overall complexity either generalize existing best ones from unconstrained or linear-constrained problems to convex-constrained ones, or improve over the best-known results on solving the same-structured problems. Furthermore, numerical experiments on linearly/quadratically constrained non-convex quadratic programs and linear-constrained robust nonlinear least squares are conducted to demonstrate the empirical efficiency of the proposed DPALM over several state-of-the art methods.
Authors: Rocco Caprio, Adam M Johansen
Abstract: We present a new framework for analysing the Expectation Maximization (EM) algorithm. Drawing on recent advances in the theory of gradient flows over Euclidean-Wasserstein spaces, we extend techniques from alternating minimization in Euclidean spaces to the EM algorithm, via its representation as coordinate-wise minimization of the free energy. In so doing, we obtain finite sample error bounds and exponential convergence of the EM algorithm under a natural generalisation of the log-Sobolev inequality. We further show that this framework naturally extends to several variants of EM, offering a unified approach for studying such algorithms.
Authors: Peilin Wu, Weiji Xie, Jiahang Cao, Hang Lai, Weinan Zhang
Abstract: Reinforcement Learning (RL) has shown its remarkable and generalizable capability in legged locomotion through sim-to-real transfer. However, while adaptive methods like domain randomization are expected to enhance policy robustness across diverse environments, they potentially compromise the policy's performance in any specific environment, leading to suboptimal real-world deployment due to the No Free Lunch theorem. To address this, we propose LoopSR, a lifelong policy adaptation framework that continuously refines RL policies in the post-deployment stage. LoopSR employs a transformer-based encoder to map real-world trajectories into a latent space and reconstruct a digital twin of the real world for further improvement. Autoencoder architecture and contrastive learning methods are adopted to enhance feature extraction of real-world dynamics. Simulation parameters for continual training are derived by combining predicted values from the decoder with retrieved parameters from a pre-collected simulation trajectory dataset. By leveraging simulated continual training, LoopSR achieves superior data efficiency compared with strong baselines, yielding eminent performance with limited data in both sim-to-sim and sim-to-real experiments. Please refer to https://peilinwu.site/looping-sim-and-real.github.io/ for videos and code.
Authors: Dwipam Katariya, Juan Manuel Origgi, Yage Wang, Thomas Caputo
Abstract: Users engage with financial services companies through multiple channels, often interacting with mobile applications, web platforms, call centers, and physical locations to service their accounts. The resulting interactions are recorded at heterogeneous temporal resolutions across these domains. This multi-channel data can be combined and encoded to create a comprehensive representation of the customer's journey for accurate intent prediction. This demands sequential learning solutions. NMT transformers achieve state-of-the-art sequential representation learning by encoding context and decoding for the next best action to represent long-range dependencies. However, three major challenges exist while combining multi-domain sequences within an encoder-decoder transformers architecture for intent prediction applications: a) aligning sequences with different sampling rates b) learning temporal dynamics across multi-variate, multi-domain sequences c) combining dynamic and static sequences. We propose an encoder-decoder transformer model to address these challenges for contextual and sequential intent prediction in financial servicing applications. Our experiments show significant improvement over the existing tabular method.
Authors: Xi Zheng, Xiangyu Chang, Ruoxi Jia, Yong Tan
Abstract: As data emerges as a vital driver of technological and economic advancements, a key challenge is accurately quantifying its value in algorithmic decision-making. The Shapley value, a well-established concept from cooperative game theory, has been widely adopted to assess the contribution of individual data sources in supervised machine learning. However, its symmetry axiom assumes all players in the cooperative game are homogeneous, which overlooks the complex structures and dependencies present in real-world datasets. To address this limitation, we extend the traditional data Shapley framework to asymmetric data Shapley, making it flexible enough to incorporate inherent structures within the datasets for structure-aware data valuation. We also introduce an efficient $k$-nearest neighbor-based algorithm for its exact computation. We demonstrate the practical applicability of our framework across various machine learning tasks and data market contexts. The code is available at: https://github.com/xzheng01/Asymmetric-Data-Shapley.
Authors: Juan Miguel Lopez Alcaraz, Nils Strodthoff
Abstract: This study investigates the feasibility of using electrocardiogram (ECG) data combined with basic patient metadata to estimate and monitor prompt laboratory abnormalities. We use the MIMIC-IV dataset to train multimodal deep learning models on ECG waveforms, demographics, biometrics, and vital signs. Our model is a structured state space classifier with late fusion for metadata. We frame the task as individual binary classifications per abnormality and evaluate performance using AUROC. The models achieve strong performance, with AUROCs above 0.70 for 24 lab values in abnormality prediction and up to 24 in abnormality forecasting, across cardiac, renal, hematological, metabolic, immunological, and coagulation categories. NTproBNP (>353 pg/mL) is best predicted (AUROC > 0.90). Other values with AUROC > 0.85 include Hemoglobin (>17.5 g/dL), Albumin (>5.2 g/dL), and Hematocrit (>51%). Our findings show ECG combined with clinical data enables prompt abnormality prediction and forecasting of lab abnormalities, offering a non-invasive, cost-effective alternative to traditional testing. This can support early intervention and enhanced patient monitoring. ECG and clinical data can help estimate and monitor abnormal lab values, potentially improving care while reducing reliance on invasive and costly procedures.
Authors: Valery Istomin, Oleg Pereziabov, Ilya Afanasyev
Abstract: Camera-captured document images often suffer from geometric distortions caused by paper deformation, perspective distortion, and lens aberrations, significantly reducing OCR accuracy. This study develops an efficient automated method for document image dewarping that balances accuracy with computational efficiency. We propose a hybrid approach combining deep learning for document detection with classical computer vision for geometry restoration. YOLOv8 performs initial document segmentation and mask generation. Subsequently, classical CV techniques construct a topological 2D grid through cubic polynomial interpolation of document boundaries, followed by image remapping to correct nonlinear distortions. A new annotated dataset and open-source framework are provided to facilitate reproducibility and further research. Experimental evaluation against state-of-the-art methods (RectiNet, DocGeoNet, DocTr++) and mobile applications (DocScan, CamScanner, TapScanner) demonstrates superior performance. Our method achieves the lowest median Character Error Rate (CER=0.0235), Levenshtein Distance (LD=27.8), and highest Jaro--Winkler similarity (JW=0.902), approaching the quality of scanned originals. The approach requires significantly fewer computational resources and memory compared to pure deep learning solutions while delivering better OCR readability and geometry restoration quality. The proposed hybrid methodology effectively restores document geometry with computational efficiency superior to existing deep learning approaches, making it suitable for resource-constrained applications while maintaining high-quality document digitization. Project page: https://github.com/HorizonParadox/DRCCBI
Authors: William Cole Nockolds, C. G. Krishnanunni, Tan Bui-Thanh, Xianxhu Tang
Abstract: Solving stiff ordinary differential equations (StODEs) requires sophisticated numerical solvers, which are often computationally expensive. In general, traditional explicit time integration schemes with restricted time step sizes are not suitable for StODEs, and one must resort to costly implicit methods. On the other hand, state-of-the-art machine learning based methods, such as Neural ODE, poorly handle the timescale separation of various elements of the solutions to StODEs, while still requiring expensive implicit/explicit integration at inference time. In this work, we propose a linear latent network (LiLaN) approach in which the dynamics in the latent space can be integrated analytically, and thus numerical integration is completely avoided. At the heart of LiLaN are the following key ideas: i) two encoder networks to encode the initial condition together with parameters of the ODE to the slope and the initial condition for the latent dynamics, respectively. Since the latent dynamics, by design, are linear, the solution can be evaluated analytically; ii) a neural network to map the physical time to latent times, one for each latent variable. Finally, iii) a decoder network to decode the latent solution to the physical solution at the corresponding physical time. We provide a universal approximation theorem for the proposed LiLaN approach, showing that it can approximate the solution of any stiff nonlinear system on a compact set to any degree of accuracy epsilon. We also show an interesting fact that the dimension of the latent dynamical system in LiLaN is independent of epsilon. Numerical results on the "Robertson Stiff Chemical Kinetics Model," "Plasma Collisional-Radiative Model," and "Allen-Cahn" and "Cahn-Hilliard" PDEs suggest that LiLaN outperformed state-of-the-art machine learning approaches for handling stiff ordinary and partial differential equations.
Authors: Kaito Takanami, Takashi Takahashi, Ayaka Sakata
Abstract: Self-distillation (SD), a technique where a model improves itself using its own predictions, has attracted attention as a simple yet powerful approach in machine learning. Despite its widespread use, the mechanisms underlying its effectiveness remain unclear. In this study, we investigate the efficacy of hyperparameter-tuned multi-stage SD with a linear classifier for binary classification on noisy Gaussian mixture data. For the analysis, we employ the replica method from statistical physics. Our findings reveal that the primary driver of SD's performance improvement is denoising through hard pseudo-labels, with the most notable gains observed in moderately sized datasets. We also identify two practical heuristics to enhance SD: early stopping that limits the number of stages, which is broadly effective, and bias parameter fixing, which helps under label imbalance. To empirically validate our theoretical findings derived from our toy model, we conduct additional experiments on CIFAR-10 classification using pretrained ResNet backbone. These results provide both theoretical and practical insights, advancing our understanding and application of SD in noisy settings.
Authors: Juan Miguel Lopez Alcaraz, Ebenezer Oloyede, David Taylor, Wilhelm Haverkamp, Nils Strodthoff
Abstract: Background: Electrocardiogram (ECG) analysis has emerged as a promising tool for detecting physiological changes linked to non-cardiac disorders. Given the close connection between cardiovascular and neurocognitive health, ECG abnormalities may be present in individuals with co-occurring neurocognitive conditions. This highlights the potential of ECG as a biomarker to improve detection, therapy monitoring, and risk stratification in patients with neurocognitive disorders, an area that remains underexplored. Methods: We aim to demonstrate the feasibility to predict neurocognitive disorders from ECG features across diverse patient populations. We utilized ECG features and demographic data to predict neurocognitive disorders defined by ICD-10 codes, focusing on dementia, delirium, and Parkinson's disease. Internal and external validations were performed using the MIMIC-IV and ECG-View datasets. Predictive performance was assessed using AUROC scores, and Shapley values were used to interpret feature contributions. Results: Significant predictive performance was observed for disorders within the neurcognitive disorders. Significantly, the disorders with the highest predictive performance is F03: Dementia, with an internal AUROC of 0.848 (95% CI: 0.848-0.848) and an external AUROC of 0.865 (0.864-0.965), followed by G30: Alzheimer's, with an internal AUROC of 0.809 (95% CI: 0.808-0.810) and an external AUROC of 0.863 (95% CI: 0.863-0.864). Feature importance analysis revealed both known and novel ECG correlates. ECGs hold promise as non-invasive, explainable biomarkers for selected neurocognitive disorders. This study demonstrates robust performance across cohorts and lays the groundwork for future clinical applications, including early detection and personalized monitoring.
Authors: Aran Nayebi
Abstract: We formalize AI alignment as a multi-objective optimization problem called $\langle M,N,\varepsilon,\delta\rangle$-agreement, in which a set of $N$ agents (including humans) must reach approximate ($\varepsilon$) agreement across $M$ candidate objectives, with probability at least $1-\delta$. Analyzing communication complexity, we prove an information-theoretic lower bound showing that once either $M$ or $N$ is large enough, no amount of computational power or rationality can avoid intrinsic alignment overheads. This establishes rigorous limits to alignment *itself*, not merely to particular methods, clarifying a "No-Free-Lunch" principle: encoding "all human values" is inherently intractable and must be managed through consensus-driven reduction or prioritization of objectives. Complementing this impossibility result, we construct explicit algorithms as achievability certificates for alignment under both unbounded and bounded rationality with noisy communication. Even in these best-case regimes, our bounded-agent and sampling analysis shows that with large task spaces ($D$) and finite samples, *reward hacking is globally inevitable*: rare high-loss states are systematically under-covered, implying scalable oversight must target safety-critical slices rather than uniform coverage. Together, these results identify fundamental complexity barriers -- tasks ($M$), agents ($N$), and state-space size ($D$) -- and offer principles for more scalable human-AI collaboration.
Authors: Laurin Lux, Alexander H. Berger, Maria Romeo Tricas, Richard Rosen, Alaa E. Fayed, Sobha Sivaprasada, Linus Kreitner, Jonas Weidner, Martin J. Menten, Daniel Rueckert, Johannes C. Paetzold
Abstract: Interpretability is crucial to enhance trust in machine learning models for medical diagnostics. However, most state-of-the-art image classifiers based on neural networks are not interpretable. As a result, clinicians often resort to known biomarkers for diagnosis, although biomarker-based classification typically performs worse than large neural networks. This work proposes a method that surpasses the performance of established machine learning models while simultaneously improving prediction interpretability for diabetic retinopathy staging from optical coherence tomography angiography (OCTA) images. Our method is based on a novel biology-informed heterogeneous graph representation that models retinal vessel segments, intercapillary areas, and the foveal avascular zone (FAZ) in a human-interpretable way. This graph representation allows us to frame diabetic retinopathy staging as a graph-level classification task, which we solve using an efficient graph neural network. We benchmark our method against well-established baselines, including classical biomarker-based classifiers, convolutional neural networks (CNNs), and vision transformers. Our model outperforms all baselines on two datasets. Crucially, we use our biology-informed graph to provide explanations of unprecedented detail. Our approach surpasses existing methods in precisely localizing and identifying critical vessels or intercapillary areas. In addition, we give informative and human-interpretable attributions to critical characteristics. Our work contributes to the development of clinical decision-support tools in ophthalmology.
Authors: Miao Zhang, Sherif Abdulatif, Benedikt Loesch, Marco Altmann, Bin Yang
Abstract: Due to the significant effort required for data collection and annotation in 3D perception tasks, mixed sample data augmentation (MSDA) has been widely studied to generate diverse training samples by mixing existing data. Recently, many MSDA techniques have been developed for point clouds, but they mainly target LiDAR data, leaving their application to radar point clouds largely unexplored. In this paper, we examine the feasibility of applying existing MSDA methods to radar point clouds and identify several challenges in adapting these techniques. These obstacles stem from the radar's irregular angular distribution, deviations from a single-sensor polar layout in multi-radar setups, and point sparsity. To address these issues, we propose Class-Aware PillarMix (CAPMix), a novel MSDA approach that applies MixUp at the pillar level in 3D point clouds, guided by class labels. Unlike methods that rely a single mix ratio to the entire sample, CAPMix assigns an independent ratio to each pillar, boosting sample diversity. To account for the density of different classes, we use class-specific distributions: for dense objects (e.g., large vehicles), we skew ratios to favor points from another sample, while for sparse objects (e.g., pedestrians), we sample more points from the original. This class-aware mixing retains critical details and enriches each sample with new information, ultimately generating more diverse training data. Experimental results demonstrate that our method not only significantly boosts performance but also outperforms existing MSDA approaches across two datasets (Bosch Street and K-Radar). We believe that this straightforward yet effective approach will spark further investigation into MSDA techniques for radar data.
Authors: Justin Le Lou\"edec, Maike Bauer, Tanja Amerstorfer, Jackie A. Davies
Abstract: Observing and forecasting coronal mass ejections (CME) in real-time is crucial due to the strong geomagnetic storms they can generate that can have a potentially damaging effect, for example, on satellites and electrical devices. With its near-real-time availability, STEREO/HI beacon data is the perfect candidate for early forecasting of CMEs. However, previous work concluded that CME arrival prediction based on beacon data could not achieve the same accuracy as with high-resolution science data due to data gaps and lower quality. We present our novel machine-learning pipeline entitled ``Beacon2Science'', bridging the gap between beacon and science data to improve CME tracking. Through this pipeline, we first enhance the quality (signal-to-noise ratio and spatial resolution) of beacon data. We then increase the time resolution of enhanced beacon images through learned interpolation to match science data's 40-minute resolution. We maximize information coherence between consecutive frames with adapted model architecture and loss functions through the different steps. The improved beacon images are comparable to science data, showing better CME visibility than the original beacon data. Furthermore, we compare CMEs tracked in beacon, enhanced beacon, and science images. The tracks extracted from enhanced beacon data are closer to those from science images, with a mean average error of $\sim 0.5 ^\circ$ of elongation compared to $1^\circ$ with original beacon data. The work presented in this paper paves the way for its application to forthcoming missions such as Vigil and PUNCH.
Authors: Scott T Steinmetz, Asmeret Naugle, Paul Schutte, Matt Sweitzer, Alex Washburne, Lisa Linville, Daniel Krofcheck, Michal Kucer, Samuel Myren
Abstract: Recent proliferation of powerful AI systems has created a strong need for capabilities that help users to calibrate trust in those systems. As AI systems grow in scale, information required to evaluate their trustworthiness becomes less accessible, presenting a growing risk of using these systems inappropriately. We propose the Trust Calibration Maturity Model (TCMM) to characterize and communicate information about AI system trustworthiness. The TCMM incorporates five dimensions of analytic maturity: Performance Characterization, Bias & Robustness Quantification, Transparency, Safety & Security, and Usability. The TCMM can be presented along with system performance information to (1) help a user to appropriately calibrate trust, (2) establish requirements and track progress, and (3) identify research needs. Here, we discuss the TCMM and demonstrate it on two target tasks: using ChatGPT for high consequence nuclear science determinations, and using PhaseNet (an ensemble of seismic models) for categorizing sources of seismic events.
Authors: Yuexin Bian, Oliver Schmidt, Yuanyuan Shi
Abstract: Energy-efficient ventilation control plays a vital role in reducing building energy consumption while ensuring occupant health and comfort. While Computational Fluid Dynamics (CFD) simulations provide detailed and physically accurate representation of indoor airflow, their high computational cost limits their use in real-time building control. In this work, we present a neural operator learning framework that combines the physical accuracy of CFD with the computational efficiency of machine learning to enable building ventilation control with the high-fidelity fluid dynamics models. Our method jointly optimizes the airflow supply rates and vent angles to reduce energy use and adhere to air quality constraints. We train an ensemble of neural operator transformer models to learn the mapping from building control actions to airflow fields using high-resolution CFD data. This learned neural operator is then embedded in an optimization-based control framework for building ventilation control. Experimental results show that our approach achieves significant energy savings compared to maximum airflow rate control, rule-based control, as well as data-driven control methods using spatially averaged CO2 prediction and deep learning based reduced order model, while consistently maintaining safe indoor air quality. These results highlight the practicality and scalability of our method in maintaining energy efficiency and indoor air quality in real-world buildings.
Authors: Zahraa Al Sahili, Ioannis Patras, Matthew Purver
Abstract: Multilingual vision-language models (VLMs) promise universal image-text retrieval, yet their social biases remain underexplored. We perform the first systematic audit of four public multilingual CLIP variants: M-CLIP, NLLB-CLIP, CAPIVARA-CLIP, and the debiased SigLIP-2, covering ten languages that differ in resource availability and morphological gender marking. Using balanced subsets of FairFace and the PATA stereotype suite in a zero-shot setting, we quantify race and gender bias and measure stereotype amplification. Contrary to the intuition that multilinguality mitigates bias, every model exhibits stronger gender skew than its English-only baseline. CAPIVARA-CLIP shows its largest biases precisely in the low-resource languages it targets, while the shared encoder of NLLB-CLIP and SigLIP-2 transfers English gender stereotypes into gender-neutral languages; loosely coupled encoders largely avoid this leakage. Although SigLIP-2 reduces agency and communion skews, it inherits -- and in caption-sparse contexts (e.g., Xhosa) amplifies -- the English anchor's crime associations. Highly gendered languages consistently magnify all bias types, yet gender-neutral languages remain vulnerable whenever cross-lingual weight sharing imports foreign stereotypes. Aggregated metrics thus mask language-specific hot spots, underscoring the need for fine-grained, language-aware bias evaluation in future multilingual VLM research.
Authors: Stefan Uhlich, Andrea Bonetti, Arun Venkitaraman, Chia-Yu Hsieh, Ya\u{g}{\i}z Gen\c{c}er, Mustafa Emre G\"ursoy, Ryoga Matsuo, Lorenzo Servadei
Abstract: We present SPICEMixer, a genetic algorithm that synthesizes circuits by directly evolving SPICE netlists. SPICEMixer operates on individual netlist lines, making it compatible with arbitrary components and subcircuits and enabling general-purpose genetic operators: crossover, mutation, and pruning, all applied directly at the netlist level. To support these operators, we normalize each netlist by enforcing consistent net naming (inputs, outputs, supplies, and internal nets) and by sorting components and nets into a fixed order, so that similar circuit structures appear at similar line positions. This normalized netlist format improves the effectiveness of crossover, mutation, and pruning. We demonstrate SPICEMixer by synthesizing standard cells (e.g., NAND2 and latch) and by designing OpAmps that meet specified targets. Across tasks, SPICEMixer matches or exceeds recent synthesis methods while requiring substantially fewer simulations.
Authors: Christian Marius Lillelund, Sanjay Kalra, Russell Greiner
Abstract: Amyotrophic lateral sclerosis (ALS) is a degenerative disorder of the motor neurons that causes progressive paralysis in patients. Current treatment options aim to prolong survival and improve quality of life. However, due to the heterogeneity of the disease, it is often difficult to determine the optimal time for potential therapies or medical interventions. In this study, we propose a novel method to predict the time until a patient with ALS experiences significant functional impairment (ALSFRS-R <= 2) for each of five common functions: speaking, swallowing, handwriting, walking, and breathing. We formulate this task as a multi-event survival problem and validate our approach in the PRO-ACT dataset (N = 3220) by training five covariate-based survival models to estimate the probability of each event over the 500 days following the baseline visit. We then predict five event-specific individual survival distributions (ISDs) for a patient, each providing an interpretable estimate of when that event is likely to occur. The results show that covariate-based models are superior to the Kaplan-Meier estimator at predicting time-to-event outcomes in the PRO-ACT dataset. Additionally, our method enables practitioners to make individual counterfactual predictions -- where certain covariates can be changed -- to estimate their effect on the predicted outcome. In this regard, we find that Riluzole has little or no impact on predicted functional decline. However, for patients with bulbar-onset ALS, our model predicts significantly shorter time-to-event estimates for loss of speech and swallowing function compared to patients with limb-onset ALS (log-rank p<0.001, Bonferroni-adjusted alpha=0.01). The proposed method can be applied to current clinical examination data to assess the risk of functional decline and thus allow more personalized treatment planning.
Authors: Rajat Rasal, Avinash Kori, Ben Glocker
Abstract: Identifiable causal representation learning seeks to uncover the true causal relationships underlying a data generation process. In medical imaging, this presents opportunities to improve the generalisability and robustness of task-specific latent features. This work introduces the concept of grouping observations to learn identifiable representations for disease classification in chest X-rays via an end-to-end framework. Our experiments demonstrate that these causal representations improve generalisability and robustness across multiple classification tasks when grouping is used to enforce invariance w.r.t race, sex, and imaging views.
Authors: Aran Nayebi
Abstract: We introduce the first complete formal solution to corrigibility in the off-switch game, with provable guarantees in multi-step, partially observed environments. Our framework consists of five *structurally separate* utility heads -- deference, switch-access preservation, truthfulness, low-impact behavior via a belief-based extension of Attainable Utility Preservation, and bounded task reward -- combined lexicographically by strict weight gaps. Theorem 1 proves exact single-round corrigibility in the partially observable off-switch game; Theorem 3 extends the guarantee to multi-step, self-spawning agents, showing that even if each head is *learned* to mean-squared error $\varepsilon$ and the planner is $\varepsilon$-sub-optimal, the probability of violating *any* safety property is bounded while still ensuring net human benefit. In contrast to Constitutional AI or RLHF/RLAIF, which merge all norms into one learned scalar, our separation makes obedience and impact-limits provably dominate even when incentives conflict. For settings where adversaries can modify the agent, we prove that deciding whether an arbitrary post-hack agent will ever violate corrigibility is undecidable by reduction to the halting problem, then carve out a finite-horizon "decidable island" where safety can be certified in randomized polynomial time and verified with privacy-preserving, constant-round zero-knowledge proofs.
Authors: Soumik Dey, Benjamin Braun, Naveen Ravipati, Hansi Wu, Binbin Li
Abstract: E-commerce sellers are advised to bid on keyphrases to boost their advertising campaigns. These keyphrases must be relevant to prevent irrelevant items from cluttering search systems and to maintain positive seller perception. It is vital that keyphrase suggestions align with seller, search and buyer judgments. Given the challenges in collecting negative feedback in these systems, LLMs have been used as a scalable proxy to human judgments. This paper presents an empirical study on a major ecommerce platform of a distillation framework involving an LLM teacher, a cross-encoder assistant and a bi-encoder Embedding Based Retrieval (EBR) student model, aimed at mitigating click-induced biases in keyphrase recommendations.
Authors: M. Adeel Ajaib, Fariha Nasir, Abdul Rehman
Abstract: We explore machine learning techniques for predicting Curie temperatures of magnetic materials using the NEMAD database. By augmenting the dataset with composition-based and domain-aware descriptors, we evaluate the performance of several machine learning models. We find that the Extra Trees Regressor delivers the best performance reaching an R^2 score of up to 0.85 $\pm$ 0.01 (cross-validated) for a balanced dataset. We employ the k-means clustering algorithm to gain insights into the performance of chemically distinct material groups. Furthermore, we perform the SHAP analysis to identify key physicochemical drivers of Curie behavior, such as average atomic number and magnetic moment. By employing explainable AI techniques, this analysis offers insights into the model's predictive behavior, thereby advancing scientific interpretability.
Authors: Daoze Zhang, Chenghan Fu, Zhanheng Nie, Jianyu Liu, Wanxian Guan, Yuan Gao, Jun Song, Pengjie Wang, Jian Xu, Bo Zheng
Abstract: With the rapid advancement of e-commerce, exploring general representations rather than task-specific ones has attracted increasing research attention. For product understanding, although existing discriminative dual-flow architectures drive progress in this field, they inherently struggle to model the many-to-one alignment between multiple images and texts of products. Therefore, we argue that generative Multimodal Large Language Models (MLLMs) hold significant potential for improving product representation learning. Nevertheless, achieving this goal still remains non-trivial due to several key challenges: the lack of multimodal and aspect-aware modeling modules in typical LLMs; the common presence of background noise in product images; and the absence of a standard benchmark for evaluation. To address these issues, we propose the first generative MLLM-based model named MOON for product representation learning. Our method (1) employs a guided Mixture-of-Experts (MoE) module for targeted modeling of multimodal and aspect-specific product content; (2) effectively detects core semantic regions in product images to mitigate the distraction and interference caused by background noise; and (3) introduces the specialized negative sampling strategy to increase the difficulty and diversity of negative samples. In addition, we release a large-scale multimodal benchmark MBE for various product understanding tasks. Experimentally, our model demonstrates competitive zero-shot performance on both our benchmark and the public dataset, showcasing strong generalization across various downstream tasks, including cross-modal retrieval, product classification, and attribute prediction. Furthermore, the case study and visualization illustrate the effectiveness of MOON for product understanding.
Authors: Alireza Fallah, Michael I. Jordan, Annie Ulichney
Abstract: Machine learning models must balance accuracy and fairness, but these goals often conflict, particularly when data come from multiple demographic groups. A useful tool for understanding this trade-off is the fairness-accuracy (FA) frontier, which characterizes the set of models that cannot be simultaneously improved in both fairness and accuracy. Prior analyses of the FA frontier provide a full characterization under the assumption of complete knowledge of population distributions -- an unrealistic ideal. We study the FA frontier in the finite-sample regime, showing how it deviates from its population counterpart and quantifying the worst-case gap between them. In particular, we derive minimax-optimal estimators that depend on the designer's knowledge of the covariate distribution. For each estimator, we characterize how finite-sample effects asymmetrically impact each group's risk, and identify optimal sample allocation strategies. Our results transform the FA frontier from a theoretical construct into a practical tool for policymakers and practitioners who must often design algorithms with limited data.
Authors: Hong Ye Tan, Subhadip Mukherjee, Junqi Tang
Abstract: Inverse problems lie at the heart of modern imaging science, with broad applications in areas such as medical imaging, remote sensing, and microscopy. Recent years have witnessed a paradigm shift in solving imaging inverse problems, where data-driven regularizers are used increasingly, leading to remarkably high-fidelity reconstruction. A particularly notable approach for data-driven regularization is to use learned image denoisers as implicit priors in iterative image reconstruction algorithms. This chapter presents a comprehensive overview of this powerful and emerging class of algorithms, commonly referred to as plug-and-play (PnP) methods. We begin by providing a brief background on image denoising and inverse problems, followed by a short review of traditional regularization strategies. We then explore how proximal splitting algorithms, such as the alternating direction method of multipliers (ADMM) and proximal gradient descent (PGD), can naturally accommodate learned denoisers in place of proximal operators, and under what conditions such replacements preserve convergence. The role of Tweedie's formula in connecting optimal Gaussian denoisers and score estimation is discussed, which lays the foundation for regularization-by-denoising (RED) and more recent diffusion-based posterior sampling methods. We discuss theoretical advances regarding the convergence of PnP algorithms, both within the RED and proximal settings, emphasizing the structural assumptions that the denoiser must satisfy for convergence, such as non-expansiveness, Lipschitz continuity, and local homogeneity. We also address practical considerations in algorithm design, including choices of denoiser architecture and acceleration strategies.
Authors: Mahmudul Islam Masum, Miad Islam
Abstract: As local AI grows in popularity, there is a critical gap between the benchmark performance of object detectors and their practical viability on consumer-grade hardware. While models like YOLOv10s promise real-time speeds, these metrics are typically achieved on high-power, desktop-class GPUs. This paper reveals that on resource-constrained systems, such as laptops with RTX 4060 GPUs, performance is not compute-bound but is instead dominated by system-level bottlenecks, as illustrated by a simple bottleneck test. To overcome this hardware-level constraint, we introduce a Two-Pass Adaptive Inference algorithm, a model-independent approach that requires no architectural changes. This study mainly focuses on adaptive inference strategies and undertakes a comparative analysis of architectural early-exit and resolution-adaptive routing, highlighting their respective trade-offs within a unified evaluation framework. The system uses a fast, low-resolution pass and only escalates to a high-resolution model pass when detection confidence is low. On a 5000-image COCO dataset, our method achieves a 1.85x speedup over a PyTorch Early-Exit baseline, with a modest mAP loss of 5.51%. This work provides a practical and reproducible blueprint for deploying high-performance, real-time AI on consumer-grade devices by shifting the focus from pure model optimization to hardware-aware inference strategies that maximize throughput.
Authors: Nivedha Sivakumar, Natalie Mackraz, Samira Khorshidi, Krishna Patel, Barry-John Theobald, Luca Zappella, Nicholas Apostoloff
Abstract: A dangerous assumption that can be made from prior work on the bias transfer hypothesis (BTH) is that biases do not transfer from pre-trained large language models (LLMs) to adapted models. We invalidate this assumption by studying the BTH in causal models under prompt adaptations, as prompting is an extremely popular and accessible adaptation strategy used in real-world applications. In contrast to prior work, we find that biases can transfer through prompting and that popular prompt-based mitigation methods do not consistently prevent biases from transferring. Specifically, the correlation between intrinsic biases and those after prompt adaptation remain moderate to strong across demographics and tasks -- for example, gender (rho >= 0.94) in co-reference resolution, and age (rho >= 0.98) and religion (rho >= 0.69) in question answering. Further, we find that biases remain strongly correlated when varying few-shot composition parameters, such as sample size, stereotypical content, occupational distribution and representational balance (rho >= 0.90). We evaluate several prompt-based debiasing strategies and find that different approaches have distinct strengths, but none consistently reduce bias transfer across models, tasks or demographics. These results demonstrate that correcting bias, and potentially improving reasoning ability, in intrinsic models may prevent propagation of biases to downstream tasks.
Authors: Karl Zhu, Dimitris Bertsimas
Abstract: Adaptive robust optimization (ARO) extends static robust optimization by allowing decisions to depend on the realized uncertainty - weakly dominating static solutions within the modeled uncertainty set. However, ARO makes previous constraints that were independent of uncertainty now dependent, making it vulnerable to additional infeasibilities when realizations fall outside the uncertainty set. This phenomenon of adaptive policies being brittle is analogous to overfitting in machine learning. To mitigate against this, we propose assigning constraint-specific uncertainty set sizes, with harder constraints given stronger probabilistic guarantees. Interpreted through the overfitting lens, this acts as regularization: tighter guarantees shrink adaptive coefficients to ensure stability, while looser ones preserve useful flexibility. This view motivates a principled approach to designing uncertainty sets that balances robustness and adaptivity.
Authors: Shijie Lian, Changti Wu, Laurence Tianruo Yang, Hang Yuan, Bin Yu, Lei Zhang, Kai Chen
Abstract: Spatial intelligence spans a rich suite of abilities, including visualising and transforming shapes, mentally rotating objects, judging relational positions and containment, and estimating numerosity. However, it still remains a critical unresolved challenge for Multimodal Large Language Models (MLLMs). To fill this gap, we propose to treat Euclidean geometry problem-solving as a surrogate task. Specifically, we meticulously constructed a curated multimodal dataset, called Euclid30K, comprising approximately 30K plane and solid geometry problems. Furthermore, to enable the model to learn and apply Euclidean principles from these geometry problems, we fine-tuned seven model variants (spanning 3--72B parameters) from the Qwen2.5VL, Qwen3VL, and RoboBrain2.0 families using Group Relative Policy Optimization (GRPO), inspiring the models to identify shapes, count, and relate entities, and perform multi-step deductive reasoning using Euclidean principles. Our experiments demonstrate that the resulting models achieve substantial zero-shot gains across four spatial reasoning benchmarks (Super-CLEVR, Omni3DBench, VSI-Bench, and MindCube) without any task-specific adaptations. Notably, after training on the Euclid30K, the mean VSI-Bench accuracy rose from 36.6\% to 41.8\% (+5.2\%), and the mean MindCube accuracy rose from 31.4\% to 38.1\% (+6.7\%). To our knowledge, this is the first systematic study showing that geometry-centric fine-tuning can confer vision-language models with broadly transferable spatial skills. Code and Euclid30K dataset can be found in \href{https://zgca-ai4edu.github.io/Euclids_Gift}{this}.
Authors: Kun Lei, Huanyu Li, Dongjie Yu, Zhenyu Wei, Lingxiao Guo, Zhennan Jiang, Ziyu Wang, Shiyu Liang, Huazhe Xu
Abstract: Real-world robotic manipulation in homes and factories demands reliability, efficiency, and robustness that approach or surpass the performance of skilled human operators. We present RL-100, a real-world reinforcement learning framework built on diffusion-based visuomotor policies. RL-100 unifies imitation and reinforcement learning under a single PPO-style objective applied within the denoising process, yielding conservative and stable policy improvements across both offline and online stages. To meet deployment latency constraints, we employ a lightweight consistency distillation procedure that compresses multi-step diffusion into a one-step controller for high-frequency control. The framework is task-, embodiment-, and representation-agnostic, and supports both single-action outputs and action-chunking control. We evaluate RL-100 on seven diverse real-robot manipulation tasks, ranging from dynamic pushing and agile bowling to pouring, cloth folding, unscrewing, and multi-stage juicing. RL-100 attains 100% success across evaluated trials, achieving 900 out of 900 successful episodes, including up to 250 out of 250 consecutive trials on one task, and matches or surpasses expert teleoperators in time-to-completion. Without retraining, a single policy attains approximately 90% zero-shot success under environmental and dynamics shifts, adapts in a few-shot regime to significant task variations (86.7%), and remains robust to aggressive human perturbations (about 95%). In a public shopping-mall deployment, the juicing robot served random customers continuously for roughly seven hours without failure. Together, these results suggest a practical path toward deployment-ready robot learning: start from human priors, align training objectives with human-grounded metrics, and reliably extend performance beyond human demonstrations.
Authors: Zia Badar
Abstract: Quantization of neural networks provides benefits of inference in less compute and memory requirements. Previous work in quantization lack two important aspects which this work provides. First almost all previous work in quantization used a non-differentiable approach and for learning; the derivative is usually set manually in backpropogation which make the learning ability of algorithm questionable, our approach is not just differentiable, we also provide proof of convergence of our approach to the optimal neural network. Second previous work in shift/logrithmic quantization either have avoided activation quantization along with weight quantization or achieved less accuracy. Learning logrithmic quantize values of form $2^n$ requires the quantization function can scale to more than 1 bit quantization which is another benifit of our quantization that it provides $n$ bits quantization as well. Our approach when tested with image classification task using imagenet dataset, resnet18 and weight quantization only achieves less than 1 percent accuracy compared to full precision accuracy while taking only 15 epochs to train using shift bit quantization and achieves comparable to SOTA approaches accuracy in both weight and activation quantization using shift bit quantization in 15 training epochs with slightly higher(only higher cpu instructions) inference cost compared to 1 bit quantization(without logrithmic quantization) and not requiring any higher precision multiplication.
Authors: Austin Xu, Xuan-Phi Nguyen, Yilun Zhou, Chien-Sheng Wu, Caiming Xiong, Shafiq Joty
Abstract: Finetuning specialized generative evaluators has emerged as a popular paradigm to meet the increasing demand for scalable evaluation during both training and test-time. However, recent work has largely focused on applying new methodology, such as reinforcement learning (RL), to training evaluators, shying away from large-scale, data-driven development. In this work, we focus on data scaling, curating a set of 2.5M samples spanning five unique evaluation tasks (pairwise, step-level, reference-free and reference-based verification, and single rating) and multiple domains focused on reasoning evaluation. With our data, we train Foundational Automatic Reasoning Evaluators (FARE), a family of 8B and 20B (with 3.6B active) parameter evaluators, with a simple iterative rejection-sampling supervised finetuning (SFT) approach. FARE-8B challenges larger specialized RL-trained evaluators and FARE-20B sets the new standard for open-source evaluators, surpassing specialized 70B+ evaluators. Beyond static benchmarks, we evaluate FARE in real-world tasks: As inference-time rerankers, FARE-20B achieves near-oracle performance on MATH. As verifiers in RL training, FARE improves the downstream RL-trained model performance by up to 14.1% vs. string-matching verifiers. When initialized from FARE, a continually-finetuned FARE-Code outperforms gpt-oss-20B by 65% on evaluating test-case quality.
Authors: Zahraa Al Sahili, Maryam Fetanat, Maimuna Nowaz, Ioannis Patras, Matthew Purver
Abstract: Text-to-image (T2I) systems lack simple, reproducible ways to evaluate how well images match prompts and how models treat social attributes. Common proxies -- face classifiers and contrastive similarity -- reward surface cues, lack calibrated abstention, and miss attributes only weakly visible (for example, religion, culture, disability). We present FairJudge, a lightweight protocol that treats instruction-following multimodal LLMs as fair judges. It scores alignment with an explanation-oriented rubric mapped to [-1, 1]; constrains judgments to a closed label set; requires evidence grounded in the visible content; and mandates abstention when cues are insufficient. Unlike CLIP-only pipelines, FairJudge yields accountable, evidence-aware decisions; unlike mitigation that alters generators, it targets evaluation fairness. We evaluate gender, race, and age on FairFace, PaTA, and FairCoT; extend to religion, culture, and disability; and assess profession correctness and alignment on IdenProf, FairCoT-Professions, and our new DIVERSIFY-Professions. We also release DIVERSIFY, a 469-image corpus of diverse, non-iconic scenes. Across datasets, judge models outperform contrastive and face-centric baselines on demographic prediction and improve mean alignment while maintaining high profession accuracy, enabling more reliable, reproducible fairness audits.
Authors: Haowei Hua (Princeton University), Hong Jiao (University of Maryland, College Park), Xinyi Wang (University of Maryland, College Park & Beijing Normal University)
Abstract: BERT and its variants are extensively explored for automated scoring. However, a limit of 512 tokens for these encoder-based models showed the deficiency in automated scoring of long essays. Thus, this research explores generative language models for automated scoring of long essays via summarization and prompting. The results revealed great improvement of scoring accuracy with QWK increased from 0.822 to 0.8878 for the Learning Agency Lab Automated Essay Scoring 2.0 dataset.
Authors: Dipal Shah (School of Natural Resources, University of Missouri, Columbia, MO, USA), Jordon Wade (Crop Protection Research & Development, Syngenta Group, Basel, Switzerland), Timothy Haithcoat (Institute for Data Science and Informatics, University of Missouri, Columbia, MO, USA), Robert Myers (School of Plant Science & Technology, University of Missouri, Columbia, MO, USA), Kelly Wilson (School of Natural Resources, University of Missouri, Columbia, MO, USA)
Abstract: Effective soil health management is crucial for sustaining agriculture, adopting ecosystem resilience, and preserving water quality. However, Missouri's diverse landscapes limit the effectiveness of broad generalized management recommendations. The lack of resolution in existing soil grouping systems necessitates data driven, site specific insights to guide tailored interventions. To address these critical challenges, a regional soil clustering framework designed to support precision soil health management strategies across the state. The methodology leveraged high resolution SSURGO dataset, explicitly processing soil properties aggregated across the 0 to 30 cm root zone. Multivariate analysis incorporating a variational autoencoder and KMeans clustering was used to group soils with similar properties. The derived clusters were validated using statistical metrics, including silhouette scores and checks against existing taxonomic units, to confirm their spatial coherence. This approach enabled us to delineate soil groups that capture textures, hydraulic properties, chemical fertility, and biological indicators unique to Missouri's diverse agroecological regions. The clustering map identified ten distinct soil health management zones. This alignment of 10 clusters was selected as optimal because it was sufficiently large to capture inherited soil patterns while remaining manageable for practical statewide application. Rooting depth limitation and saturated hydraulic conductivity emerged as principal variables driving soil differentiation. Each management zone is defined by a unique combination of clay, organic matter, pH, and available water capacity. This framework bridges sophisticated data analysis with actionable, site targeted recommendations, enabling conservation planners, and agronomists to optimize management practices and enhance resource efficiency statewide.
Authors: Fei Yang
Abstract: Human prowess in intuitive physics remains unmatched by machines. To bridge this gap, we argue for a fundamental shift towards brain-inspired computational principles. This paper introduces the Spatiotemporal Relational Neural Network (SRNN), a model that establishes a unified neural representation for object attributes, relations, and timeline, with computations governed by a Hebbian ``Fire Together, Wire Together'' mechanism across dedicated \textit{What} and \textit{How} pathways. This unified representation is directly used to generate structured linguistic descriptions of the visual scene, bridging perception and language within a shared neural substrate. On the CLEVRER benchmark, SRNN achieves competitive performance, thereby confirming its capability to represent essential spatiotemporal relations from the visual stream. Cognitive ablation analysis further reveals a benchmark bias, outlining a path for a more holistic evaluation. Finally, the white-box nature of SRNN enables precise pinpointing of error root causes. Our work provides a proof-of-concept that confirms the viability of translating key principles of biological intelligence into engineered systems for intuitive physics understanding in constrained environments.
Authors: Pawe{\l} Liskowski, Benjamin Han, Paritosh Aggarwal, Bowei Chen, Boxin Jiang, Nitish Jindal, Zihan Li, Aaron Lin, Kyle Schmaus, Jay Tayade, Weicheng Zhao, Anupam Datta, Nathan Wiegand, Dimitris Tsirogiannis
Abstract: Snowflake's Cortex AISQL is a production SQL engine that integrates native semantic operations directly into SQL. This integration allows users to write declarative queries that combine relational operations with semantic reasoning, enabling them to query both structured and unstructured data effortlessly. However, making semantic operations efficient at production scale poses fundamental challenges. Semantic operations are more expensive than traditional SQL operations, possess distinct latency and throughput characteristics, and their cost and selectivity are unknown during query compilation. Furthermore, existing query engines are not designed to optimize semantic operations. The AISQL query execution engine addresses these challenges through three novel techniques informed by production deployment data from Snowflake customers. First, AI-aware query optimization treats AI inference cost as a first-class optimization objective, reasoning about large language model (LLM) cost directly during query planning to achieve 2-8$\times$ speedups. Second, adaptive model cascades reduce inference costs by routing most rows through a fast proxy model while escalating uncertain cases to a powerful oracle model, achieving 2-6$\times$ speedups while maintaining 90-95% of oracle model quality. Third, semantic join query rewriting lowers the quadratic time complexity of join operations to linear through reformulation as multi-label classification tasks, achieving 15-70$\times$ speedups with often improved prediction quality. AISQL is deployed in production at Snowflake, where it powers diverse customer workloads across analytics, search, and content understanding.
Authors: Yuxin Bai, Aranyak Acharyya, Ashwin De Silva, Zeyu Shen, James Hassett, Joshua T. Vogelstein
Abstract: Optimal control of the future is the next frontier for AI. Current approaches to this problem are typically rooted in either reinforcement learning (RL). While powerful, this learning framework is mathematically distinct from supervised learning, which has been the main workhorse for the recent achievements in AI. Moreover, RL typically operates in a stationary environment with episodic resets, limiting its utility to more realistic settings. Here, we extend supervised learning to address learning to control in non-stationary, reset-free environments. Using this framework, called ''Prospective Learning with Control (PL+C)'', we prove that under certain fairly general assumptions, empirical risk minimization (ERM) asymptotically achieves the Bayes optimal policy. We then consider a specific instance of prospective learning with control, foraging -- which is a canonical task for any mobile agent -- be it natural or artificial. We illustrate that modern RL algorithms fail to learn in these non-stationary reset-free environments, and even with modifications, they are orders of magnitude less efficient than our prospective foraging agents.
Authors: Minrui Luo, Weihang Xu, Xiang Gao, Maryam Fazel, Simon Shaolei Du
Abstract: Gradient descent dynamics on the deep matrix factorization problem is extensively studied as a simplified theoretical model for deep neural networks. Although the convergence theory for two-layer matrix factorization is well-established, no global convergence guarantee for general deep matrix factorization under random initialization has been established to date. To address this gap, we provide a polynomial-time global convergence guarantee for randomly initialized gradient descent on four-layer matrix factorization, given certain conditions on the target matrix and a standard balanced regularization term. Our analysis employs new techniques to show saddle-avoidance properties of gradient decent dynamics, and extends previous theories to characterize the change in eigenvalues of layer weights.
Authors: Ao Xu, Han Zhao, Weihao Cui, Quan Chen, Yukang Chen, Shulai Zhang, Shuang Chen, Jiemin Jiang, Zhibin Yu, Minyi Guo
Abstract: Large language models (LLMs) are increasingly deployed under the Model-as-a-Service (MaaS) paradigm. To meet stringent quality-of-service (QoS) requirements, existing LLM serving systems disaggregate the prefill and decode phases of inference. However, decode instances often experience low GPU utilization due to their memory-bound nature and insufficient batching in dynamic workloads, leaving compute resources underutilized. We introduce Harli, a serving system that improves GPU utilization by co-locating parameter-efficient finetuning (PEFT) tasks with LLM decode instances. PEFT tasks are compute-bound and memory-efficient, making them ideal candidates for safe co-location. Specifically, Harli addresses key challenges--limited memory and unpredictable interference--using three components: a unified memory allocator for runtime memory reuse, a two-stage latency predictor for decode latency modeling, and a QoS-guaranteed throughput-maximizing scheduler for throughput maximization. Experimental results show that Harli improves the finetune throughput by 46.2% on average (up to 92.0%) over state-of-the-art serving systems, while maintaining strict QoS guarantees for inference decode.
Authors: Rasmus F. Orsoe, Stephan Meighen-Berger, Jeffrey Lazar, Jorge Prado, Ivan Mozun-Mateo, Aske Rosted, Philip Weigel, Arturo Llorente Anaya
Abstract: Neutrino telescopes are large-scale detectors designed to observe Cherenkov radiation produced from neutrino interactions in water or ice. They exist to identify extraterrestrial neutrino sources and to probe fundamental questions pertaining to the elusive neutrino itself. A central challenge common across neutrino telescopes is to solve a series of inverse problems known as event reconstruction, which seeks to resolve properties of the incident neutrino, based on the detected Cherenkov light. In recent times, significant efforts have been made in adapting advances from deep learning research to event reconstruction, as such techniques provide several benefits over traditional methods. While a large degree of similarity in reconstruction needs and low-level data exists, cross-experimental collaboration has been hindered by a lack of diverse open-source datasets for comparing methods. We present NuBench, an open benchmark for deep learning-based event reconstruction in neutrino telescopes. NuBench comprises seven large-scale simulated datasets containing nearly 130 million charged- and neutral-current muon-neutrino interactions spanning 10 GeV to 100 TeV, generated across six detector geometries inspired by existing and proposed experiments. These datasets provide pulse- and event-level information suitable for developing and comparing machine-learning reconstruction methods in both water and ice environments. Using NuBench, we evaluate four reconstruction algorithms - ParticleNeT and DynEdge, both actively used within the KM3NeT and IceCube collaborations, respectively, along with GRIT and DeepIce - on up to five core tasks: energy and direction reconstruction, topology classification, interaction vertex prediction, and inelasticity estimation.
Authors: Abdul Rehman, Iqra Rasool, Ayisha Imran, Mohsen Ali, Waqas Sultani
Abstract: Digital hematopathology requires cell-level analysis across diverse disease categories, including malignant disorders (e.g., leukemia), infectious conditions (e.g., malaria), and non-malignant red blood cell disorders (e.g., sickle cell disease). Whether single-task, vision-language, WSI-optimized, or single-cell hematology models, these approaches share a key limitation, they cannot provide unified, multi-task, multi-modal reasoning across the complexities of digital hematopathology. To overcome these limitations, we propose Uni-Hema, a multi-task, unified model for digital hematopathology integrating detection, classification, segmentation, morphology prediction, and reasoning across multiple diseases. Uni-Hema leverages 46 publicly available datasets, encompassing over 700K images and 21K question-answer pairs, and is built upon Hema-Former, a multimodal module that bridges visual and textual representations at the hierarchy level for the different tasks (detection, classification, segmentation, morphology, mask language modeling and visual question answer) at different granularity. Extensive experiments demonstrate that Uni-Hema achieves comparable or superior performance to train on a single-task and single dataset models, across diverse hematological tasks, while providing interpretable, morphologically relevant insights at the single-cell level. Our framework establishes a new standard for multi-task and multi-modal digital hematopathology. The code will be made publicly available.
Authors: Pattaraphon Kenny Wongchamcharoen, Paul Glasserman
Abstract: Large language models (LLMs) are increasingly used in finance and economics, where prompt-based attempts against look-ahead bias implicitly assume that models understand chronology. We test this fundamental question with a series of chronological ordering tasks with increasing complexities over facts the model already knows from pre-training. Our tasks cover (1) chronological ordering, (2) conditional sorting (filter, then order), and (3) anachronism detection. We evaluate GPT-4.1, Claude-3.7 Sonnet, with and without Extended Thinking (ET), and GPT-5 across multiple reasoning-effort settings. Across models, Exact match rate drops sharply as sequences lengthen even while rank correlations stay high as LLMs largely preserve local order but struggle to maintain a single globally consistent timeline. In conditional sorting, most failures stem from the filtering step rather than the ordering step, but GPT-5 and Claude-3.7 Sonnet with Extended Thinking outshine normal models significantly. Lastly, anachronism detection is found to be the easiest task for the LLMs but performance still declines with increasingly overlapping timelines or entities. Overall, our main contribution is showing that allocating explicit reasoning budget helps with chronological ordering with GPT-5 at medium/high reasoning effort achieving flawless ordering at all lengths and perfect conditional sorting (both self-filtered and given-subset), whereas low/minimal effort degrades with longer lists, mirroring earlier models. Our findings delineate limits of current LLMs on chronological tasks, providing insights into task complexity, and demonstrate scenarios in which reasoning helps. These patterns are important for the real-time application of LLMs in finance. We release all code and evaluation templates to support full reproducibility.
Authors: Alessio Pellegrino, Jacopo Mauro
Abstract: One of the long-standing goals in optimisation and constraint programming is to describe a problem in natural language and automatically obtain an executable, efficient model. Large language models appear to bring this vision closer, showing impressive results in automatically generating models for classical benchmarks. However, much of this apparent success may derive from data contamination rather than genuine reasoning: many standard CP problems are likely included in the training data of these models. To examine this hypothesis, we systematically rephrased and perturbed a set of well-known CSPLib problems to preserve their structure while modifying their context and introducing misleading elements. We then compared the models produced by three representative LLMs across original and modified descriptions. Our qualitative analysis shows that while LLMs can produce syntactically valid and semantically plausible models, their performance drops sharply under contextual and linguistic variation, revealing shallow understanding and sensitivity to wording.