4219 papers
This paper introduces Bradley-Terry Policy Optimization (BTPO), a novel framework that derives a consistent Monte Carlo gradient estimator to effectively train large language models with chain-of-thought reasoning on non-verifiable pairwise preference tasks, overcoming the limitations of existing heuristic RL approaches.
This paper introduces AffPCL, a personalized collaborative learning framework that utilizes affinity-based variance reduction to automatically adapt to unknown heterogeneity among agents, achieving sample complexity improvements that seamlessly interpolate between independent learning and linear speedup without requiring prior knowledge of system similarity.
FALCON addresses the spatial reasoning limitations of existing 2D-based vision-language-action models by leveraging spatial foundation models to inject rich 3D geometric priors directly into the action head, achieving state-of-the-art performance across diverse simulation and real-world tasks without requiring architectural changes or specialized sensors.
The paper proposes GraphKeeper, a novel framework for Graph Domain-Incremental Learning that addresses catastrophic forgetting through knowledge disentanglement and deviation-free preservation, achieving state-of-the-art performance across multiple graph domains while remaining compatible with various graph foundation models.
This paper proposes a structured matrix scaling approach for multi-class calibration that leverages theoretical insights from logistic regression, combined with structured regularization and robust optimization, to effectively manage the bias-variance tradeoff and achieve substantial performance gains over existing methods while providing an open-source implementation.
This paper proposes LTSV, a lightweight and efficient method for valuing time series data in foundation models by leveraging in-context finetuning and temporal block aggregation to overcome the computational bottlenecks and temporal dependency limitations of traditional approaches.
This paper introduces a novel in-context learning approach using Time-Series Foundation Models (TSFMs) to classify bearing health status from vibration data without fine-tuning, enabling scalable, zero-shot maintenance solutions across varying operational conditions.
This paper proposes "Periodic Asynchrony," a framework that accelerates LLM reinforcement learning by decoupling inference and training into a provably on-policy asynchronous pipeline, achieving a 3- to 5-fold throughput improvement on NPU platforms without sacrificing accuracy.
This paper introduces Decoder-DeepONet (DDON), a novel interpretable operator-learning model that significantly outperforms previous neural network and classical methods in reconstructing electric field profiles from EFISH signals by offering superior accuracy, generalizability, and robustness to incomplete data while identifying optimal sampling windows.
This paper introduces ELERAG, an enhanced Retrieval-Augmented Generation system that integrates Wikidata-based Entity Linking and a hybrid re-ranking strategy to significantly improve factual accuracy in Italian educational question-answering, particularly outperforming standard methods in domain-specific contexts while demonstrating the importance of domain-adapted strategies.
This paper introduces SAGFM, a robust Graph Foundation Model framework that enhances domain-adaptive representations and generalization by integrating structure-aware semantic augmentation, an information bottleneck mechanism, and expert adaptive routing to effectively mitigate domain noise, structural perturbations, and adversarial attacks.
ADHint is a novel reinforcement learning framework that enhances reasoning capabilities and generalization by integrating sample difficulty priors to adaptively schedule hint ratios and employing consistency-based gradient modulation with rollout difficulty posteriors to stabilize learning and prevent destructive imitation.
This paper introduces Directional Textual Inversion (DTI), a method that improves personalized text-to-image generation by constraining learned token embeddings to a fixed magnitude and optimizing only their direction on a hypersphere, thereby preventing norm inflation, enhancing text fidelity, and enabling smooth semantic interpolation between concepts.
This study demonstrates that while the multichannel linear descriptors-based block field method (MLD-BFM) achieves the highest accuracy in decoding five finger-joint degrees of freedom from HD sEMG, its performance is not statistically superior to conventional time-domain features, though it significantly outperforms dimensionality reduction methods, highlighting the critical importance of preserving spatial resolution in high-density recordings.
This paper introduces EMFusion, a conditional multivariate diffusion-based framework that leverages a residual U-Net with cross-attention and imputation-based sampling to provide accurate, uncertainty-quantified, frequency-selective electromagnetic field forecasts for wireless network planning, significantly outperforming existing baseline models.
This study introduces the enhanced WT-RDF+ framework, which leverages machine learning-assisted parameter tuning to overcome amplitude accuracy limitations in reconstructing Radial Distribution Functions for amorphous Ge-Se and Ag-Ge-Se systems, thereby outperforming standard ML benchmarks even with limited training data.
This paper identifies a systematic mismatch in activation updates during gradient descent, proposing that correcting this issue through first-principles derivation not only reinterprets the role of normalization but also yields a new, empirically superior alternative called "PatchNorm" that challenges conventional affine-based approaches.
This paper rigorously proves that incorporating local updates into the distributed DIGing algorithm can provably accelerate convergence for a broad class of objective functions, demonstrating that two local updates are sufficient to achieve maximal improvement without requiring reduced step sizes.
This paper introduces Bayesian Generative Modeling (BGM), a unified framework that leverages a stochastic iterative Bayesian updating algorithm to learn a single generative model capable of performing arbitrary conditional inference with principled uncertainty quantification, without requiring retraining for different conditioning structures.
This paper presents an automated system using LLM-powered web research agents to generate and resolve diverse, real-world forecasting questions at scale, demonstrating high-quality question creation and resolution rates that surpass human-curated platforms while effectively evaluating and improving AI forecasting performance.