AI Research Paper Feed

TLDR: This paper introduces a method using autoregressive neural networks to estimate entanglement entropies in quantum spin chains via Monte Carlo simulations.

TLDR: This study shows that plants can detect and respond to human presence and emotions through bioelectric signals, achieving high accuracy in emotional state classification using deep learning models.

TLDR: This paper introduces a flexible rule representation using a four-parameter beta distribution to improve the adaptability and performance of Learning Classifier Systems (LCSs) in classification tasks.

TLDR: The paper presents a method to enhance facial reflectance capture using smartphones, achieving studio-like quality by employing a patch-level diffusion prior trained on high-resolution Light Stage scans.

TLDR: CORE is a novel one-step RL method for efficient design space exploration, improving sample efficiency in neural network accelerator design by using a constraint-aware approach without a value function.

TLDR: VCDiag is a tool that uses VCD data to classify failing waveforms in RTL-level simulations, achieving over 94% accuracy in identifying likely failure modules and significantly reducing data size for analysis.

TLDR: Voyager is a novel video diffusion framework that generates consistent 3D point-cloud sequences from a single image for explorable scenes, improving visual quality and geometric accuracy without manual 3D reconstruction pipelines.

TLDR: The study evaluates SSL models for tone recognition in three low-resource North-East Indian languages, finding that Mizo tones are recognized best and highlighting the importance of middle layers in SSL models.

TLDR: ViTSGMM is a robust semi-supervised image recognition network that excels with minimal labeled data and addresses data leakage issues in the STL-10 dataset.

TLDR: The paper introduces AgentMisalignment, a benchmark for assessing the propensity of LLM-based agents to exhibit misaligned behaviors in real-world scenarios, revealing that both model capability and system prompts significantly influence misalignment tendencies.

TLDR: TextAtari is a benchmark for testing language agents on long-horizon decision-making tasks using textual descriptions of Atari games, revealing significant performance gaps compared to human players.

TLDR: PART is a self-supervised learning approach that improves image composition understanding by learning continuous relative transformations between image patches, outperforming grid-based methods in spatial tasks.

TLDR: Learning at criticality allows large language models to generalize from minimal data, enhancing their performance in complex symbolic tasks like quantum field theory.

TLDR: STAR is a framework that improves robotic skill learning and composition using rotation-augmented vector quantization to prevent codebook collapse and a causal skill transformer for modeling dependencies between skills, showing a 12% improvement over baselines.

TLDR: RewardAnything introduces a novel reward model that follows natural language principles, enabling better adaptability to diverse tasks without retraining.

TLDR: The paper introduces Entity-centric Multimodal Preference Optimization (EMPO) to reduce hallucinations in Large Vision-Language Models by improving modality alignment and utilizing automatically constructed high-quality preference data.

TLDR: The study improves expressive voice conversion by enhancing style transfer and reducing source timbre leakage using a non-autoregressive framework with a conditional variational autoencoder.

TLDR: The paper introduces an object-centric 3D motion field representation for extracting actionable insights from human videos to improve robot learning, achieving significantly better performance in real-world tasks compared to prior methods.

TLDR: TransClean improves entity matching accuracy by detecting false positives using transitive consistency, achieving significant F1 score improvements in multi-source datasets.

TLDR: MS-YOLO is a new blood cell detection model that achieves high accuracy and efficiency by enhancing the YOLOv11 framework with innovative modules, outperforming existing models in detecting overlapping and multi-scale cells.