Interface-Aware Neural Newton Preconditioning for Robust Cohesive Zone Model Simulations
Neural Newton preconditioning methods for cohesive zone model simulations in composite structures.
Neural Newton preconditioning methods for cohesive zone model simulations in composite structures.
Study on Generated Contents Enrichment for conditional image generation with explicit scene representations.
Image augmentation technique (ForAug) that factorizes training images into foreground/background to mitigate compositional biases in image classification.
Optimization algorithm for low-rank problems with sparse simplex constraints using multiplicative updates on oblique manifolds.
SE(3)-equivariant graph neural networks framework for inferring protein conformational ensembles from sparse experimental data.
Interpretable multimodal deep transformation models combining neural networks with statistical methods for stroke outcome prediction with Grad-CAM explainability.
Theoretical analysis of variance assumptions in stochastic gradient optimization algorithms from classical to modern perspectives.
Pattern Embedded Neural Networks (PENNs) for multivariate regression with missing covariates, combining imputation with indicator observation networks.
Energy-efficient deep learning model for 4-stage sleep classification using single-lead ECG signals for home-based monitoring.
Efficient sparse attention kernel implementation for LLMs, providing alternative to Native Sparse Attention with improved performance on long-context tasks.
Cognitive science study on human reinforcement learning showing how compressed reward functions transfer from working memory to long-term memory.
Machine learning approach using learned regularizers to improve signal recovery from amplitude-only measurements beyond theoretical recovery limits.
Test adequacy metric (Clotho) for measuring LLM input quality before inference to enable task-specific prompt testing without ground truth outputs.
Theoretical analysis of identifying root cause nodes in cyclic causal graphs by tracing outlier propagation through structural equations.
Large-scale dataset of 400K visual chain-of-thought reasoning examples for training multimodal large language models on spatially grounded reasoning tasks.
Vision-language model approach for zero-shot distracted driver detection using double decoupling to separate appearance variations from behavior cues.
Dynamic sparse attention mechanism for video diffusion transformers to reduce quadratic complexity of self-attention in long-sequence generation tasks.
Discrete diffusion framework for handwritten mathematical expression recognition using symbol-aware masked diffusion to improve consistency and reduce exposure bias.
MMLoP low-rank prompting method for efficient vision-language model adaptation with millions fewer parameters than state-of-the-art.
DICE-RL framework using RL as distribution contraction operator to finetune pretrained diffusion/flow-based robot policies from online feedback.
IRIS benchmark: 240 4K real-world videos for unsupervised physical parameter estimation and governing-equation identification.
Surrogate modeling framework for stochastic differential equations using path-space observable error bounds for efficient simulation.
LLM-powered pipeline for automated extraction and structuring of materials science data from unstructured scientific literature.
Geometric and dynamical analysis of attractor basins and storage limits in kernel Hopfield networks; real-world image embedding evaluation.
HeadsUp scalable feed-forward method for 3D Gaussian head reconstruction from multi-view captures using encoder-decoder architecture.
Neural network emulators for real-time tokamak plasma shape control via virtual circuits; replaces expensive numerical computation.
Non-asymptotic Wasserstein contraction analysis for coordinate ascent variational inference algorithms.
Physics-informed neural network for subsurface mineral prospectivity modeling incorporating Darcy flow and advection-diffusion.
Theoretical analysis of queue peak dynamics in stochastic networks under nonstationary arrival conditions.
Physics-informed review of deep learning statistical properties including neural scaling laws from classical statistics perspective.
FeLoG distributed graph embedding framework with feedback loop mechanism for billion-scale graphs; applications include GraphRAG.
Taxonomic strategy retrieval approach to mitigate compounding failures in multi-step agentic tasks; addresses problem drift in persuasion agents.
Mathematical framework for matrix-gated blended score estimation in Ornstein-Uhlenbeck diffusion sampling; theoretical contribution.
Large-scale simulation framework for physical adversarial acoustic attacks against voice-controlled AI systems; addresses scaling challenges.
DataComp-VLM benchmark with 160 datasets for systematically evaluating vision-language model training data curation strategies.
Study of evaluation inconsistencies in diffusion LLM decoding strategies; identifies reproducible sources of bias in efficiency benchmarking.
HASTE hierarchical multi-agent system for ML engineering that accumulates cross-competition knowledge via LLM-driven abstraction across three scope tiers.
Academic perspective on AI's role in scientific research infrastructure and risks of homogenization.
Open-source auditable sandbox for recording and monitoring AI coding agent actions, behaviors, and file modifications.
PostgreSQL procedural language extension enabling PHP functions and triggers in database.
UI design tool for real devices with export of layout specifications and prompts for AI code generation.
HN discussion thread asking about local LLM deployments in organizations, hardware choices, and operational challenges.
Open-source DOCX editor and SDK for rendering, editing, and automating Word documents in browsers and AI agent workflows.
Founder perspectives on scaling AI adoption across organizations and engineering teams.
DocETL enables processing large data collections with LLMs using natural language operations and map-reduce patterns.
Open-source Windows desktop app for AI workspace. Supports cloud APIs, local LLMs, tool agents, and workflow orchestration with governance tracking.
Case studies on using TimescaleDB open-source database for energy and carbon data analysis at scale.
Bare title with no content.
Open-source duplicate code detector for AI-generated code using structural fingerprinting instead of text matching.
News item on Kling AI fundraising round and valuation.