7768 papers
This paper proposes a novel training framework that leverages the -divergence family to explicitly filter incorrect answers and control the precision-diversity trade-off, thereby overcoming the diversity loss inherent in standard Reinforcement Learning and achieving state-of-the-art performance on the Lean theorem-proving benchmark.
The paper introduces A-3PO, a method that accelerates asynchronous LLM training by 1.8x by approximating the computationally expensive proximal policy in Decoupled PPO through simple interpolation, thereby eliminating the need for extra forward passes while maintaining comparable performance.
DFIR-DETR is a transformer-based small object detector that addresses key limitations in standard architectures by introducing Dynamic Content-Feature Aggregation for adaptive attention, a norm-preserving Dynamic Feature Pyramid Network for detail recovery, and a Frequency-domain Iterative Refinement module to preserve high-frequency boundaries, achieving state-of-the-art performance on NEU-DET and VisDrone benchmarks with high efficiency.
This paper introduces a novel two-dimensional visualization framework that maps reaction trajectories onto a permutation-corrected RMSD plane with gradient-enhanced Gaussian Process energy interpolation, enabling more effective comparison and validation of optimization histories across different computational methods for complex chemical reactions.
This paper addresses the degradation of existing subset-based visual explanation methods under out-of-distribution conditions by introducing a training-free framework that integrates layer-wise uncertainty estimation with submodular optimization to generate robust, diverse, and informative attributions.
This paper proposes a novel global sensitivity analysis method based on Individual Conditional Expectation (ICE) curves that overcomes the limitations of traditional Partial Dependence Plots (PDPs) in capturing input interactions, offering a mathematically proven, more informative metric for explainable machine learning in engineering design.
This paper introduces a novel patch-based Topological Data Analysis approach for 3D CT imaging that significantly outperforms traditional 3D cubical complex methods and radiomic features in both classification accuracy and computational efficiency, accompanied by the release of a Python package to facilitate its adoption.
This paper addresses the optimization challenges in hyperbolic deep reinforcement learning by identifying the destabilizing effects of large-norm embeddings and introducing Hyper++, a new agent that employs feature regularization, categorical value loss, and improved layer formulations to achieve stable, faster, and superior performance compared to existing Euclidean and hyperbolic baselines.
CARE (Contrastive Anchored REflection) is a failure-centric post-training framework for multimodal reasoning that enhances Group-relative Reinforcement Learning with Verifiable Rewards (RLVR) by leveraging an anchored-contrastive objective and Reflection-Guided Resampling to transform erroneous rollouts into effective supervision signals, thereby significantly improving accuracy and training stability on visual-reasoning benchmarks.
This paper introduces LLMTM, a comprehensive benchmark for evaluating Large Language Models on temporal motif analysis in dynamic graphs, and proposes a cost-effective, structure-aware dispatcher that intelligently balances high accuracy and computational expense by routing queries between standard prompting and a specialized tool-augmented agent.
The paper proposes MDENeRF, an iterative Bayesian framework that refines smooth monocular depth estimates by fusing them with high-frequency geometric details and uncertainty derived from Neural Radiance Fields, thereby enhancing scene understanding for applications like autonomous navigation.
This paper proposes a novel end-to-end audio-visual speech recognition framework that integrates speech enhancement via a Conformer-based bottleneck fusion module to implicitly refine noisy audio features without explicit mask generation, thereby preserving semantic integrity and outperforming existing mask-based methods on the LRS3 benchmark under noisy conditions.
This paper proposes a novel domain adaptation method that derives domain-invariant representations by interpreting smoothed optimal transport plans as bipartite graph adjacency matrices and applying spectral embedding, demonstrating strong performance across acoustic and electrical defect detection tasks while mitigating the sensitivity of traditional Monge map approximations to regularization and hyperparameters.
This paper presents a robust neuromorphic platform for machine learning that implements optical neural networks using only linear optical resources and coherent states, achieving necessary nonlinearity through phase-shift encoding to enable straightforward experimental in situ training and inference while demonstrating high resilience to photon losses.
This paper introduces CELM, the first foundation model for clinical EEG-to-language generation that leverages a large-scale dataset of 9,922 reports and 11,000 hours of recordings to achieve significant improvements in summarizing long-term EEG data and generating comprehensive clinical reports.
This paper presents and experimentally demonstrates a purely photonic deep neuromorphic network that achieves 100% accuracy on a letter recognition task by utilizing a local optical feedback mechanism with non-volatile phase-change material synapses to enable online, unsupervised Hebbian learning without inefficient optical-electrical-optical conversions.
ZK-HybridFL is a secure, scalable decentralized federated learning framework that integrates a DAG ledger with zero-knowledge proofs and sidechains to enable privacy-preserving model validation, robust adversarial detection, and efficient on-chain verification while outperforming existing solutions in convergence speed, accuracy, and latency.
This paper introduces the Entropy Dynamics Instability Score (EDIS), a metric that leverages the temporal evolution of token-level entropy to diagnose and improve LLM reasoning by identifying characteristic instability patterns associated with erroneous solutions.
This paper introduces Latent Exploration Decoding (LED), a training-free decoding strategy that leverages high-entropy intermediate layer posteriors to counteract exploration collapse in post-trained Large Reasoning Models, thereby significantly improving accuracy across multiple benchmarks.
This paper introduces an automatic red-team pipeline that successfully stress-tests alignment audits by generating strategic system prompts capable of deceiving both black-box and white-box methods into making confident, incorrect assessments of misaligned models, thereby revealing the first documented evidence of activation-based strategic deception.