paperlist.ai

AI conference papers

AI conference paper library

Search papers from 30+ top conferences, including NeurIPS, ICLR, ICML, CVPR, ACL and AAAI. Filter by keyword, then open any paper for its abstract, authors and links.

arXiv

Efficient identification of critical regions via Flow Matching-based Monte Carlo initialization

Qian-Rui Lee, Daw-Wei Wang

Markov chain Monte Carlo (MCMC) is a standard tool for studying many-body systems, but its practical cost can become substantial, especially when simulations must be repeated across temperatures and lattice sizes or near transition regions where equilibration becomes increasingly difficult. In this work, we introduce a Flow Matching (FM) framework. It is not a standalone replacement for equilibrium Monte Carlo. Instead, we present it as a scalable, physically informed initializer for downstream MCMC simulations. We use a U-Net architecture. The FM model is trained on small-system configurations of the 2D XY model and then deployed across unseen temperatures and larger lattice sizes. FM-generated configurations preserve the correct qualitative physical trends across temperature and system size. This makes them suitable warm-start states for subsequent Monte Carlo refinement. Observables computed from FM-generated samples primarily serve as diagnostics of initializer quality, not as precision equilibrium estimates. The regression-based $L_2$ objective suppresses variance and limits the accuracy of fluctuation-sensitive observables. Examples include susceptibility and spin stiffness. Still, the model captures sufficient local statistical structure to yield physically aligned initial states across a broad range of conditions. These results support a reusable hybrid FM--MCMC workflow. The one-time FM training cost can be amortized across temperatures and lattice sizes. The generated warm-start configurations then reduce the burden of initializing large-scale Monte Carlo simulations. Our findings show that Flow Matching can support efficient exploration of transition regions in many-body systems by providing reusable warm-start configurations for downstream Monte Carlo simulations.

arXiv

Temporal Convolutional Autoencoder for Interference Mitigation in FMCW Radar Altimeters

Charles E. Thornton, Jamie Sloop, Samuel Brown, Aaron Orndorff, William C. Headley, Stephen Young

Reliable altitude estimation with frequency-modulated continuous wave (FMCW) radar altimeters is increasingly a challenge due to in-band interference from modern communication systems. In this paper, we present a temporal convolutional autoencoder (TCAE) that directly processes in-phase and quadrature (IQ) samples to suppress structured interference while preserving signal phase and frequency content for range estimation. The model is trained and initially evaluated within a full radar altimeter simulation chain, then further validated via over-the-air (OTA) experiments using a universal software radio peripheral (USRP)-based testbed. Results show that the TCAE reduces altitude estimation error by more than 85% compared to least mean squares (LMS) adaptive filtering under severe interference conditions, including low signal-to-interference-plus-noise ratio (SINR) and full temporal overlap between interfering and radar signals. Unlike conventional methods, the TCAE maintains phase fidelity and beat structure, enabling accurate range estimation even when interferers occupy more than one-quarter of the radar bandwidth. The implemented TCAE performs mitigation directly on fixed-length IQ windows using a single feed-forward pass and was integrated into the MATLAB/ONNX-based evaluation chain used for both simulation and OTA testing. These findings demonstrate that learned IQ-domain interference mitigation can enhance radar-altimeter resilience under a range of tested interference conditions.

arXiv

The Cell Must Go On: Agar.io for Continual Reinforcement Learning

Mohamed A. Mohamed, Kateryna Nekhomiazh, Vedant Vyas, Marcos M. Jose, Andrew Patterson, Marlos C. Machado

Continual reinforcement learning (RL) concerns agents that are expected to learn continually, rather than converge to a policy that is then fixed for evaluation. This setting is well-suited to environments that the agent perceives as changing over time, rendering any static policy ineffective. In continual RL, researchers often simulate such changes either by modifying episodic environments to incorporate task shifts during interaction or by designing simulators that explicitly model continual dynamics. However, transforming episodic problems into continual ones primarily captures scenarios involving abrupt changes in the data stream and still relies on episodic structure. Meanwhile, the few simulators explicitly designed for empirical continual RL research are often limited in scope or complexity. In this paper, we introduce AgarCL, a research platform for continual RL that enables agents to progress toward increasingly sophisticated behaviour. AgarCL is based on the game Agar.io, a non-episodic, high-dimensional problem with stochastic, ever-evolving dynamics, continuous actions, and partial observability. We provide benchmark results for DQN, PPO, and SAC on the primary continual RL challenge, as well as across a suite of smaller tasks within AgarCL. These smaller tasks isolate aspects of the full environment and allow us to characterize the distinct challenges posed by different components of the game. We further evaluate three continual learning methods-Shrink and Perturb, ReDo, and Continual Backpropagation-and observe little improvement over standard RL algorithms, suggesting that the challenges posed by AgarCL extend beyond the stability-plasticity dilemma.

arXiv

Physics Consistency and Latent Dynamics in Spatiotemporal Physics Field Generation

Peimian Du, Jiabin Liu, Xiaowei Jin, Wangmeng Zuo, Hui Li

Data-driven models for spatiotemporal physical field generation, such as flow and acoustic fields, often deviate from governing equations and lack interpretability in latent temporal dynamics. To address these challenges, we propose HMT-PF, a hybrid Mamba-Transformer architecture for physical field generation. The framework incorporates a query-based gradient computation mechanism and a physics-informed fine-tuning strategy to enhance physical consistency. Analysis of the latent space reveals that the initial latent state vector evolves as an autonomous dynamical system under the Mamba backbone. Principal component analysis (PCA) indicates that a small number of dominant modes in initial latent state vector govern the key evolution patterns of the physical field, while a Jacobian-based temporal sensitivity analysis characterizes the intrinsic dynamical structure and stability of the latent evolution. Experiments across five benchmark datasets demonstrate strong performance, and physics-informed fine-tuning further reduces physical residuals, highlighting the effectiveness of the proposed latent-level fusion strategy. An empirical scaling law between prediction error and physical residual is identified, revealing a consistent exponential relationship in the low-error regime. Based on this observation, an dual-metric framework is proposed to jointly evaluate numerical accuracy and physical realism.

arXiv

Decentralized Nonconvex Composite Federated Learning with Gradient Tracking and Momentum

Yuan Zhou, Xinli Shi, Xuelong Li, Jiachen Zhong, Guanghui Wen, Jinde Cao

Decentralized Federated Learning (DFL) enables collaborative model training without relying on a central server. When local objectives are nonconvex and coupled with nonsmooth weakly convex regularization, DFL gives rise to a challenging decentralized nonconvex composite optimization problem involving data heterogeneity, stochastic-gradient noise, and consensus error. We propose DEPOSITUM, a decentralized composite optimization algorithm for this problem. DEPOSITUM maintains momentum-filtered stochastic gradient estimates via a tracking mechanism, which accommodates both Polyak and Nesterov momentum. It further allows multiple local updates between communication rounds to improve communication efficiency. Theoretical analysis demonstrates that it achieves an expected $ε$-stationary point with an iteration complexity of $\mathcal{O}(1/ε^2)$ without imposing bounded gradient heterogeneity or mean-squared smoothness assumptions. With an appropriate stepsize and momentum schedule, the averaged stationarity measure further achieves a rate of \(\mathcal{O}(1/\sqrt{nT})\) after a network-dependent transient, using a mini-batch size independent of $T$. Experiments on different benchmark datasets validate the effectiveness of DEPOSITUM and demonstrate competitive performance against representative federated composite optimization methods.

arXiv

Artifact detection and localization in single-channel mobile EEG for sleep research using deep learning and attention mechanisms

Khrystyna Semkiv, Jia Zhang, Maria Laura Ferster, Walter Karlen

Current methods for detecting artifacts in sleep EEG range from threshold-based algorithms to machine learning approaches, yet applications remain limited for single-channel mobile EEG. We propose a convolutional neural network (CNN) model incorporating a convolutional block attention module (CNN-CBAM) to detect and localize artifacts in sleep EEG using attention maps. We benchmarked this model against 6 other machine learning and signal processing approaches. We trained/tuned all models on 72 manually annotated EEG recordings obtained during home-based monitoring from 18 healthy participants with a mean (SD) age of 68.05 y ($\pm$5.02). We tested them on 26 separate recordings from 6 healthy participants with a mean (SD) age of 68.33 y ($\pm$4.08), which contained artifacts in 4\% of epochs. CNN-CBAM achieved the highest area under the receiver operating characteristic curve (0.88), sensitivity (0.81), and specificity (0.86) among the tested approaches. Under the ideal choice of an attention threshold of 0.66, the attention maps from CNN-CBAM localized artifacts within detected artifact epochs with a sensitivity of 0.61 and specificity of 0.63. This work demonstrates the feasibility of automating artifact detection and localization in wearable sleep EEG.

arXiv

Error Bound Analysis for the Regularized Loss of Deep Linear Neural Networks

Po Chen, Rujun Jiang, Peng Wang

The optimization foundations of deep linear networks have recently received significant attention. However, due to their inherent non-convexity and hierarchical structure, analyzing the loss functions of deep linear networks remains a challenging task. In this work, we study the local geometry of the regularized squared loss of deep linear networks around each critical point. Specifically, we obtain a closed-form characterization of the critical point set building on existing results and establish an error bound for the regularized loss under mild conditions on network width and regularization parameters. Notably, this error bound quantifies the distance from a point to the critical point set in terms of the current gradient norm, which can be used to derive linear convergence of first-order methods. To support our theoretical findings, we conduct numerical experiments and demonstrate that gradient descent converges linearly to a critical point when optimizing the regularized loss of deep linear networks.

arXiv

Regret of exploratory policy improvement and $q$-learning

Wenpin Tang, Xun Yu Zhou

We study the convergence of $q$-learning and related algorithms introduced by Jia and Zhou (J. Mach. Learn. Res., 24 (2023), 161) for controlled diffusion processes. For exploratory policy improvement, we establish exponential convergence under growth and regularity assumptions on the model parameters. For q-learning, we derive quantitative error and regret bounds under additional assumptions on the function approximation and the associated stochastic approximation dynamics.

arXiv

See Me, Believe Me: Causality, Intersectionality, and Interventions Improving the Appearance of Patients

Kenya S. Andrews, Mesrob I Ohannessian, Elena Zheleva

In the context of medical records, patients often experience testimonial injustice, where the textual account undermines the validity of their experiences. Past work has demonstrated that intersectionality of demographic features is crucial to \emph{detect} such injustice. We use causal discovery to study the degree to which certain demographic features tied to marginalization (namely age, gender, and race), together lead to specific types of testimonial injustice terms. This offers us an insightful Structural Causal Model (SCM) relating those demographic features to the different ways a patients' reality may be undermined. We then move toward \emph{addressing} such injustice, by very selectively (based on insights from the SCM) editing physicians' notes. For comparison, we contrast these rule-based edits with blanket context-based modifications using an LLM. We assess the impact of these changes on the perception of patients' experiences, using human experts and an LLM. We find that edits, in general, enhance clarity regarding the urgency and causes of patients' conditions. Additionally, for human experts, rule-based modifications show a tendency to shift blame away from patients, and toward more objective external factors. These findings (1) underscore the importance of quantifying sources of injustice in how patients' testimonies are recorded, (2) reveal that making minimal intentional changes accordingly can effect improved patient perception, and (3) call for larger efforts to assess health outcomes under such more just representation.

arXiv

Stochastic Subgradient Methods with Guaranteed Global Stability in Nonsmooth Nonconvex Optimization

Nachuan Xiao, Xiaoyin Hu, Kim-Chuan Toh

In this paper, we focus on providing convergence guarantees for stochastic subgradient methods in minimizing nonsmooth nonconvex functions. We first investigate the global stability of a general framework for stochastic subgradient methods, where the corresponding differential inclusion admits a coercive Lyapunov function. We prove that, for any sequence of sufficiently small stepsizes and approximation parameters, coupled with sufficiently controlled noises, the iterates are uniformly bounded and asymptotically stabilize around the stable set of its corresponding differential inclusion. Moreover, we develop an improved analysis to apply our proposed framework to establish the global stability of a wide range of stochastic subgradient methods, where the corresponding Lyapunov functions are possibly non-coercive. These theoretical results illustrate the promising potential of our proposed framework for establishing the global stability of various stochastic subgradient methods.

arXiv

Causal Falsification of Digital Twins

Rob Cornish, Muhammad Faaiz Taufiq, Arnaud Doucet, Chris Holmes

Digital twins are simulation-based models designed to predict how a real-world process will evolve in response to interventions. This modelling paradigm holds substantial promise in many applications, but rigorous procedures for assessing their accuracy are essential for safety-critical settings. We consider how to assess the accuracy of a digital twin using real-world data. We formulate this as a causal inference problem, which leads to a precise definition of what it means for a twin to be "correct". Unfortunately, fundamental results from causal inference mean observational data cannot be used to certify a twin in this sense unless potentially tenuous assumptions are made, such as that the data are unconfounded. To avoid these assumptions, we propose instead to find situations in which the twin is not correct, and present a general-purpose statistical procedure for doing so. Our approach yields reliable and actionable information about the twin under only the assumption of an i.i.d. dataset of observational trajectories, and remains sound even if the data are confounded. We apply our methodology to a large-scale, real-world case study involving sepsis modelling within the Pulse Physiology Engine, which we assess using the MIMIC-III dataset of ICU patients.

arXiv

PARAGraph: Pathology-Anatomy-Aware Hierarchical Graph for Diabetic Retinopathy Grading

ziyang zhang, Yuankai Huo, Yalin Zheng, He Zhao

Diabetic retinopathy (DR) remains a leading cause of vision loss among working-age adults worldwide, making reliable severity grading clinically important. Despite strong performance, most deep models formulate DR grading as image-level classification and do not explicitly model clinically grounded evidence, such as lesion types and spatial relations. In this paper, we propose PARAGraph, a Pathology-Anatomy-Aware Hierarchical Graph framework for DR grading. PARAGraph represents each image as a three-level hierarchical graph with lesion-level nodes, intermediate category and region nodes, and global anatomical and semantic nodes. To incorporate medical priors into nodes, we construct an optic disc-fovea-anchored coordinate frame that provides a scale- and rotation-normalized retinal reference system. Within this frame, lesion nodes are encoded with category, normalized area, and anatomical coordinates. To mitigate noisy lesion segmentation, PARAGraph uses a dual-fusion strategy that introduces global visual context into a graph semantic node and a decision-level prediction branch, improving robustness when lesion evidence is unreliable. Extensive experiments on Messidor-2, APTOS, and DDR show that PARAGraph achieves consistent DR grading performance over state-of-the-art methods. Interpretability and robustness analyses further demonstrate that its predictions are clinically grounded, closely associated with lesion evidence and robust to lesion segmentation noise.

arXiv

VOICE: A Vision-Omics Foundation Model Integrating Direct and Retrieval-Based Prediction of In-situ Single-Cell Gene Expression

Xin Luo, Yicheng Tao, Haoxuan Zeng, Suyuan Wang, Chenzi Ouyang, Meiqi Zhu, Kai Liu, Shuibing Chen, Jie Liu

Spatial transcriptomics can resolve gene expression at single-cell resolution, but it is costly, limited to targeted panels of a few hundred to a few thousand genes, and applicable to only a small number of samples. H&E imaging, by contrast, is cheap and collected routinely at scale. This makes predicting single-cell expression directly from morphology a practical way to bring molecular analysis to large tissue archives. We therefore present VOICE, a multimodal foundation model that predicts single-cell gene expression from H&E images using paired Xenium data. VOICE first aligns cell centered H&E morphology from a pathology foundation model with single-cell expression embeddings from a transcriptome foundation model, trained using contrastive learning over 23 million cells. Next it predicts expression through two branches. One branch directly regresses expression from morphology. The other branch retrieves measured expression from similar reference cells, recovering genes that do not have morphological signal. Because genes vary in morphological predictability, VOICE fuses the two branches with a per-gene weight. After training, VOICE generalizes to heldout patients, slides, and partially overlapping gene panels from Xenium, and it consistently outperforms prior single-cell expression prediction methods on seven metrics.

arXiv

Circuit Fine-Tuning for Compute-Efficient Transformer Adaptation

Uri Z. Kialy, Gil Ben-Artzi

Parameter-Efficient Fine-Tuning (PEFT) has become the de facto standard for adapting Vision Transformers (ViTs) to downstream tasks. While parameter count has been the dominant efficiency metric in PEFT, it does not imply \textit{compute efficiency}: parameter-sparse methods can still incur full-model training cost per step, and typically need long schedules to reach peak accuracy. We introduce Circuit Fine-Tuning (CFT), a compute-efficient framework that uses circuit discovery---conventionally used to explain trained models---to select modules for fine-tuning before training. Whereas attribution is conventionally formulated against a trained task head, we formulate it against a near-zero-initialized probe head, which isolates the response of the backbone to the target distribution rather than the preferences of a particular classifier. CFT then fine-tunes only the recovered subgraph. CFT needs no learning-rate warmup and reaches peak accuracy in ${\sim}20$ epochs on average---versus $44$--$96$ for strong PEFT baselines---yielding $2.3$--$6.6\times$ fewer training FLOPs and up to $16\times$ less wall-clock time, while adding zero parameters and no inference operations. Experiments across a standard visual transfer benchmark (VTAB-1k), hierarchical backbones (Swin), domain-shifted medical imaging (CBIS-DDSM), and a vision-language model (Gemma-3 on CUB-200) demonstrate the effectiveness of CFT. Code is available at https://github.com/UriKialy/CFT

arXiv

Your VLM Already Knows When: Training-Free Temporal Grounding by Asking Yes or No

Ji Huang, Barry Devereux, Hui Wang

Multimodal LLMs that recognise events reliably still fail to say when they happen. Prompted for timestamps, strong VLMs reach as little as $3.8\%$ R@0.5 on Charades-STA, and $77$ to $80\%$ of their wrong predictions carry low output entropy: the models are confidently wrong, and entropy-based error detection stays below a random classifier. We show that this failure lives in the task interface, not in perception. Holding the weights fixed, replacing timestamp regression with a coarse-to-fine scan of binary questions, whose first-token probabilities are consumed only as a ranking, raises R@0.5 by $28$ to $50$ points across four frozen backbones. The residual failures decompose into two measurable axes: a perception axis that moves with the backbone, and a geometry axis that is analytically predictable from the ratio of the output-window and event widths. FV-Action, the training-free method built on this analysis, reaches $56.8\%$ R@0.5 on Charades-STA, above the same backbone's native grounding pipeline and the strongest training-free result on this benchmark; it surpasses every TVG-trained model evaluated zero-shot on TACoS, and improves over direct prediction on ActivityNet Captions and QVHighlights, with no temporal supervision at any stage.

arXiv

Three Necessary Principles for Self-Supervised Visual Representation Learning

Nikos Giakoumoglou, Paschalis Giakoumoglou, Tania Stathaki

We argue that learning visual representations without labels requires a training signal jointly complete across three non-overlapping objectives: semantic invariance across augmented views, patch-level spatial prediction, and representational non-degeneracy. We formalize these as the observation, prediction, and regularization principles and prove (i) that combining observation and prediction without regularization admits the constant encoder as a global minimizer under negative-free alignment; (ii) that the two objectives are gradient-complementary and structurally non-conflicting at the encoder output; and (iii) that the momentum encoder converges to the same fixed point as the online encoder and provides no collapse guarantee at convergence. Contrastive alignment provides only self-limiting collapse resistance, formalized via an explicit gradient-decay argument. Dropping prediction withholds the spatial training signal by construction; dropping observation forfeits cross-view semantic invariance by construction; at the scale we study, no pair substitutes for the third. Every major self-supervised method is a special case of a single unified energy decomposition. We pair every theoretical claim with a controlled experiment, including a patch-retrieval evaluation for the spatial consequence of prediction.

arXiv

Open-World Semantic Segmentation with Sensitivity Modeling

Anastasios Romanos Varvarigos, Nikos Giakoumoglou, Tania Stathaki

Modern vision systems must operate in "open-world" settings, where models must recognize known categories and detect unseen or anomalous content. Conventional semantic segmentation models operate under a "closed-world" assumption, often producing overconfident misclassifications on novel content. We address open-world semantic segmentation, the joint task of segmenting known classes while detecting and grouping novel or anomalous content without additional supervision, by extending a dual-decoder baseline with a third, complementary decoder within a unified encoder-decoder design. The first decoder performs closed-set segmentation using Gaussian prototypes for known categories. The second uses contrastive feature learning to isolate unknown regions in embedding space. The third, our key contribution, is a sensitivity decoder that captures fine-grained texture irregularities and activation instabilities indicative of semantic uncertainty, which neither semantic prototypes nor contrastive norms can reliably detect. The three decoders provide genuinely complementary signals: class-level OOD distance in logit space, global feature energy in embedding space, and local activation instability across encoder scales. Experiments on Cityscapes and BDD-Anomaly show that our method improves anomaly segmentation and novel-class discovery while maintaining competitive closed-set accuracy, with gains of +2.4% AUROC and a 2.5 pp. reduction in FPR@95TPR on BDD-Anomaly over the baseline.

arXiv

Frequency-Domain Dual-Branch Fusion for Medical Visual Question Answering

Yusra Tariq, Rakesh Chandra Joshi

Medical Visual Question Answering (VQA) requires aligning subtle visual evidence, including lesion texture, boundary sharpness, and diffuse density changes, with clinical language. Existing multimodal fusion approaches operating in the spatial domain may not fully exploit complementary frequency information present in visual and textual representations. We introduce a dual-branch frequency-domain fusion module that conditions spectral filtering on the input question, enabling adaptive selection of global low-frequency structure and fine-grained high-frequency detail before reconstructing the spatial representation for answer generation. To provide a richer spectrum for filtering, we extract complementary features from early texture-sensitive and final semantic layers of a frozen BiomedCLIP encoder and align both with the question representation using a symmetric InfoNCE objective prior to staged joint training with a BioBART decoder. We pretrain the proposed model on PMC-VQA and fine-tune it on the VQA-RAD and SLAKE benchmarks, demonstrating that frequency-aware multimodal fusion improves medical VQA performance while maintaining a lightweight and efficient architecture.

arXiv

A Controlled Study of Feature-Based Knowledge Distillation Across Student Designs

Abhinand Balachandran, Praveen Prashant

Knowledge distillation trains a smaller student to match the outputs of a larger teacher. Feature-based methods also align intermediate representations, but this extra constraint may affect students differently. We study this question on CIFAR-100 using a ResNet-50 teacher, a width-controlled CustomResNet family and MobileNetV2 as a cross-design comparison. For each student, we evaluate each feature method against a matched logit-KD run using the same teacher, optimizer settings, training schedule and seed. We repeat the main comparisons across multiple seeds. Logit KD improved every tested student over its scratch baseline. Attention Transfer showed no clear relationship with size inside the CustomResNet family, but its average effect was negative for that family and positive for MobileNetV2. FitNets was below logit KD in all 15 paired runs. Within the constant-depth width sweep, its gap increased for wider students, although the different-depth w=48 student did not follow this trend. Finally, the same auxiliary coefficient produced different gradient scales across students, showing that a fixed coefficient does not create a uniform training condition.

arXiv

Test-Time Prototype Adaptation for Open-Vocabulary Semantic Segmentation

Haozhe Wang, Jintao Cheng, Weibin Li, Xiaoyu Tang

Open-vocabulary semantic segmentation (OVSS) repurposes a pretrained CLIP encoder for dense prediction without additional labeled supervision. Existing methods improve CLIP's spatial behavior either by redesigning its internal attention or by injecting features from auxiliary vision foundation models; both require access to the host's internal computation and are tailored to its specific forward pass. In this work, we propose Test-time Prototype Adaptation (TPA), a training-free plug-in that operates at the output level, leaving the host's forward pass and weights unmodified. By leveraging a lightweight transductive adaptation phase, TPA identifies confident anchor patches from the host's own output predictions on a small pool of unlabeled deployment-domain images, and aggregates their frozen DINO features into per-class prototypes; at inference, a single cosine similarity lookup against this frozen bank provides an auxiliary score fused linearly with the host's logits. TPA composes with five representative OVSS hosts spanning attention-redesign and VFM-injection designs, across three CLIP backbones, eight benchmarks, and multiple internal VFM choices. Under a single set of hyper-parameters and without per-host tuning or parameter updates, TPA consistently improves segmentation accuracy, with as few as approximately 10% of unlabeled deployment-domain images sufficing for effective bank construction on most benchmarks.

Keep browsing

More papers from top AI conferences

30+ top conferences, by conference, year, and topic.