Yan Liu

The Physical Mapping Guard (PMG), grounded in the software engineering principle of Separation of Concerns, revokes verification authority from the agent, forcing semantic intents to be evaluated by an external, deterministic Semantic-to-Physical (S2P) mapping engine.

LabDex reveals how a structured approach to task taxonomy can enhance the training and evaluation of robots in complex laboratory settings.

Morphology-derived tumor states reveal critical progression information that traditional diagnostic labels overlook, linking spatial histopathology to multi-omics insights.

Uncertainty-aware predictions can cut binding affinity prediction errors by 25%, revolutionizing trust in AI-driven drug discovery.

TempoWave reveals that rethinking numerical embeddings can unlock significant improvements in LLM forecasting performance.

GLACIER achieves high predictive accuracy while drastically cutting down the computational costs typically associated with multimodal molecular property prediction.

Ada reveals the intricate decision-making processes of software engineering agents, transforming raw trajectory data into actionable insights about their behavior.

Despite its smaller size, Qwen3-8B, when guided by a novel "Think Thrice Before You Speak" framework, achieves superior performance to GPT-5 in predicting desires, beliefs, and persuasive strategies within persuasive dialogues.

Instead of passively transcribing doctor-patient dialogues, this system actively models what's known, what's missing, and what questions to ask next, paving the way for more intelligent EMR systems.