Newest

FAR-DPO is a feasibility-aware and robust direct preference optimization framework that enhances cyclic peptide design for drug discovery by aligning generative models with structural and biophysical constraints, improving success rates on benchmarks like PepGLAD and PepFlow.

This paper proposes Answer-Level Trust Selection (ATS), a post-hoc, model-agnostic framework for assessing the reliability of individual predictions from vision-language models in quantitative physical reasoning tasks.

MileGPO proposes a method for credit assignment in long-horizon LLM agents using milestone inference with local evidence, achieving state-of-the-art performance on ALFWorld and WebShop benchmarks.

This paper investigates unsupervised anomaly detection using flow matching on tabular data, focusing on contaminated training sets and comparing different scoring methods for robustness.

The paper analyzes the delayed effects of minibatch perturbations in AdamW by modeling it as a finite-horizon input-state-output system, revealing how optimizer states influence training dynamics.

This paper audits step-level credit assignment in LLM agents using causal ground truth from executed replay, finding that common credit signals fail to identify causally important steps better than chance, with implications for training methods.

The paper proposes TUP, a method for BoN-style distillation via rank-based classification that truncates low-ranked completions and upweights high-ranked ones to improve alignment efficiency and performance.

RecPFN introduces a prior-fitted network for in-context learning in sequential recommendations, pretrained on synthetic clickstream data to achieve state-of-the-art zero-shot performance across benchmarks.

This paper introduces a locally tokenized generative model for robust watermarking in multivariate time-series data, addressing reliability issues under post-editing attacks by using bounded temporal neighborhoods for token recovery.

SAGE-XGBoost is a machine learning framework that uses spatially augmented graph embeddings and data augmentation to enhance natural hazard susceptibility mapping under data scarcity, demonstrating superior performance over conventional models.

FleetSieve introduces a decision-critical profiling method for SLO-aware LLM fleet configuration that optimizes resource allocation by reducing unnecessary measurements, achieving efficiency gains over uniform profiling.

The article introduces a rationally enriched Chebyshev trunk for DeepONet surrogate models, enhancing accuracy in simulating high-Péclet transport problems with thin boundary layers.

Introduces DeltaML-Bench, a benchmark for evaluating machine learning agents on real-world research repositories, showing that ARG scaffolding significantly improves success rates over standard methods.

This paper identifies an error in a widely used time-uniform self-normalized concentration inequality for discounted least squares in non-stationary bandit problems, provides counterexamples and lower bounds, and offers corrected inequalities.

This study systematically evaluates physics-informed neural network techniques for fluid dynamics, showing that combining periodic activations with causal weighting improves performance on the Navier-Stokes vortex shedding benchmark, while further additions cause degradation due to nonlinear interactions.

This paper analyzes a single-loop, entropy-regularized Natural Actor-Critic algorithm and proves accelerated convergence rates for the unregularized objective in stochastic and deterministic regimes under linear function approximation.

This paper explores optimization landscapes in complex neural networks using Kähler geometry and information manifolds, providing theoretical guarantees on descent paths and analyzing effects of Calabi-Yau metrics.

This paper presents a two-stage transformer framework for predicting the risk and timing of acute exacerbations of COPD using raw waveform data from home ventilators, demonstrating improved performance over baseline models.

DraftFM is a foundation model for predicting draft picks in Magic: The Gathering, enabling accurate day-zero drafting for new sets by leveraging card features and behavioral data. It demonstrated strong performance on held-out expansions and successfully forecasted an unreleased set.

This paper proposes MeanFlow-Transfer (MF-T) and Continuous Adversarial MeanFlow (CAMF) to unify the adaptation and acceleration of pretrained diffusion and flow models, enabling high-quality few-step generation on new domains with limited data.