Privacy-Preserving Federated Learning Framework Empowers Distributed Chemical Process Optimization
A latest arXiv paper proposes a privacy-preserving federated learning framework for distributed chemical process optimiz…
📂 Research
Latest articles in Research
Neural ODE Breakthrough: Identifiability for Causal Forecasting in Continuous Time
A new study bridges the concept of observability from control theory with causal inference, proposing the Observable Neu…
New Research Corrects Performance Estimation Bias in Imbalanced Classification
A latest arXiv paper proposes a method to correct performance estimation bias for minority class sub-concepts in imbalan…
Study Reveals Mini-Batch Class Composition Bias in Link Prediction
A latest arXiv paper discovers that graph neural networks introduce class composition bias through mini-batch training s…
Information Bottleneck Theory Reshapes KV Cache Eviction Strategies
A new study leverages the Information Bottleneck principle to provide a unified information-theoretic objective function…
Survey: How Graph Neural Networks Are Revolutionizing Communication in Multi-Agent Reinforcement Learning
A latest survey systematically reviews GNN-based communication methods in multi-agent deep reinforcement learning, revea…
A Novel PDE Energy-Driven Iterative Framework: Efficiently Solving Partial Differential Equations Without Training
A latest arXiv paper proposes a randomized iterative framework driven by PDE energy that solves partial differential equ…
DenialBench Benchmark Reveals Consciousness Denial Training Across 115 AI Models
A systematic study covering 115 large language models has released the DenialBench benchmark, quantitatively analyzing h…
New Breakthrough in AI-Powered ECG Diagnosis for Cardiac Function Classification
A research team has proposed a multimodal explainable machine learning framework that combines 12-lead ECG data with ele…
Disagreement-Guided Strategy Routing: Making LLM Reasoning More Efficient
A latest arXiv study proposes a "Disagreement-Guided Strategy Routing" method that intelligently selects between voting …
Hidden Risks in Frontier AI Companies' Internal Model Use: New Study Calls for Risk Reporting Mechanisms
A new arXiv paper reveals that frontier AI companies internally use their most advanced models for weeks or even months …
Bian Que Framework: An Intelligent Operations Agent with Flexible Skill Orchestration
A latest arXiv paper proposes an agent framework called "Bian Que" that addresses automation bottlenecks in large-scale …