6611 papers
This paper proposes Rank-factorized Implicit Neural Bias (RIB), a novel positional bias mechanism that enables the use of hardware-efficient FlashAttention in Super-Resolution Transformers, allowing for significantly larger window sizes and training patches that achieve state-of-the-art performance (35.63 dB PSNR) while reducing training and inference times by 2.1 and 2.9, respectively.
This paper introduces an efficient framework for heterogeneous decentralized diffusion models that enables experts to train with mixed objectives (DDPM and Flow Matching) and reduced resource requirements, achieving a 16x decrease in compute and 14x reduction in data compared to prior approaches while improving image quality and diversity.
This paper proposes a framework that fine-tunes pretrained generative models to directly sample within complex, structured feasible regions, achieving a novel balance between strict constraint satisfaction and high-quality sample realism for safety-critical applications like robotics and autonomous driving.
This paper identifies that applying Group Relative Policy Optimization (GRPO) to diffusion language models causes reward collapse due to noisy importance ratio estimates and formulation mismatches, and proposes StableDRL, a reformulated algorithm featuring unconditional clipping and self-normalization to stabilize training and prevent policy drift.
The paper introduces DIRECTER, a novel activation steering method that dynamically modulates steering strength through a plausibility-guided decoding loop and layer sensitivity analysis to enhance LLM instruction-following accuracy while preventing the oversteering that typically degrades text quality.
This paper introduces ProtAlign, a multi-objective preference alignment framework that fine-tunes pretrained inverse folding models to simultaneously optimize diverse developability properties like solubility and thermostability while preserving structural designability, resulting in the enhanced MoMPNN model for practical protein sequence design.
This paper proposes the Latent Autoencoder Ensemble Kalman Filter (LAE-EnKF), a novel data assimilation method that learns a stable, linear state-space model in a latent space to overcome the performance limitations of standard EnKF on strongly nonlinear and chaotic systems while maintaining computational efficiency.
This paper proposes a hybrid quantum-classical autoencoder and variational autoencoder framework utilizing Quantum Implicit Neural Representations (QINR) to achieve stable, high-quality image reconstruction and generation with enhanced diversity and sharp details compared to existing quantum generative models.
This paper proposes ICD3, an interpretable and robust approach for detecting concept drift in imbalanced streaming data by employing multi-distribution-granular search to identify small concepts and training independent One-Cluster Classifiers for each, thereby overcoming the masking effect of dominant large clusters.
This paper proposes and validates a multi-level explainability framework demonstrating that SHAP explanations for Alzheimer's disease diagnostic and prognostic models are robust, stable, and consistent across different disease stages and prediction tasks, thereby enhancing their reliability for clinical adoption.
This paper proposes a diversity-aware adaptive collocation method for Physics-Informed Neural Networks that formulates point selection as a sparse QUBO optimization problem on a kNN graph to efficiently construct hybrid coreset subsets, thereby reducing training redundancy and overhead while improving accuracy on PDEs with shock formation.
This study presents a machine learning framework for predicting steady-state flow through porous media, demonstrating that the Fourier Neural Operator (FNO) outperforms convolutional autoencoders and U-Nets by achieving high accuracy, significant computational speedups over traditional CFD, and mesh-invariant properties ideal for topology optimization.
This paper proposes a meta-learning framework combined with graph convolutional neural networks to enable rapid adaptation of traffic flow predictions to unseen network disruptions and demand patterns, achieving high accuracy (R² ≈ 0.85) without requiring extensive training data covering all possible scenarios.
This paper demonstrates that symbolic machine learning can effectively predict failures in chemical processes, such as ethylene oxidation, by generating interpretable, rule-based models that outperform traditional black-box methods while addressing the scarcity of real-world failure data through simulator-generated examples.
This paper argues that neural representations possess an inherent gauge freedom under invertible linear transformations, rendering standard similarity metrics like cosine similarity coordinate-dependent and unstable, thereby necessitating analysis based on gauge-invariant quantities or explicitly defined canonical coordinates.
This paper proposes HGT-Scheduler, a deep reinforcement learning framework that utilizes Heterogeneous Graph Transformers to explicitly model the distinct edge semantics of the Job Shop Scheduling Problem, thereby outperforming homogeneous graph baselines on benchmark instances by capturing type-specific relational patterns through edge-type-dependent attention mechanisms.
SpatialMAGIC is a hybrid framework that integrates graph diffusion and transformer-based spatial self-attention to effectively impute sparse and noisy spatial transcriptomics data, thereby enhancing clustering accuracy, improving gene detection, and preserving biological interpretability across multiple high-resolution platforms.
The paper introduces **xaitimesynth**, an open-source Python package that streamlines the evaluation of time series attribution methods by providing a reusable infrastructure for generating synthetic datasets with known ground truth masks and calculating standard localization metrics.
To address the critical data scarcity of extreme rare weather events that hinders robust machine learning models, this paper proposes a physics-informed diffusion model based on Context-UNet that generates physically consistent, multi-spectral synthetic satellite imagery conditioned on key atmospheric parameters, thereby effectively mitigating extreme class imbalance and enhancing operational weather detection algorithms.
This paper introduces Optimistic Policy Regularization (OPR), a lightweight mechanism that preserves and reinforces historically successful trajectories to prevent premature convergence, thereby significantly improving sample efficiency and final performance in deep reinforcement learning across Atari and cyber-defense benchmarks.