About
Research & Development
Targeted Data Removal: A New Strategy for Unlearning in Large Language Models
TLDR: MAPE-Unlearn is a new method for machine unlearning in Transformer models that efficiently removes specific data influences. Unlike previous methods, it uses learnable...
Research & Development
Simulating Justice: AI Agents Transform Chinese Court Training and Legal Practice
TLDR: SimCourt is a groundbreaking framework that uses advanced LLM-based agents to simulate the entire process of Chinese criminal trials. It replicates all five...
Research & Development
Enhancing Speech Clarity: A Dual-Microphone Approach Adapts to Noise Levels
TLDR: A new multi-modal framework called BAF-Net combines body-conduction microphone signals (BMS) and acoustic microphone signals (AMS) for superior speech enhancement. It uses separate,...
Research & Development
Mapping Legal Logic: How Knowledge Graphs Uncover Court Reasoning
TLDR: A new research paper introduces a method to build Legal Knowledge Graphs (LKGs) from Japanese court judgments. These graphs explicitly map the complex...
Research & Development
A Hybrid Approach to Self-Supervised Vision Transformers: DinoTwins Unveiled
TLDR: DinoTwins combines DINO's self-distillation and Barlow Twins' redundancy reduction to train Vision Transformers without extensive labeled data. This hybrid model achieves comparable classification...
Research & Development
Activation Transport Operators: A New Lens on Transformer Internal Dynamics
TLDR: Activation Transport Operators (ATOs) are linear maps that predict how features move between layers in transformer models. They help determine if a feature...