Houde Liu

MASTE achieves zero-shot Aspect Sentiment Triplet Extraction with a multi-agent approach that outperforms traditional LLM methods, even without labeled data.

Many robotic policies that seem successful in manipulation tasks actually compromise safety, with SoftVTBench revealing a stark contrast between goal completion and physical safety metrics.

Transforming unstructured expert videos into structured rewards, STDR boosts RL agent performance in robotic manipulation by improving both efficiency and robustness.

Autonomous exploration by an LLM agent dramatically outperforms both rigid retrieval workflows and supervised fine-tuning for temporal knowledge graph question answering, achieving state-of-the-art results in a zero-shot setting.