Qiang Li

Agentic hybrid RAG outperforms existing methods in retrieving and synthesizing evidence for muon collider research, setting a new standard for scientific question answering in high-energy physics.

Medical VQA models can now reason more reliably thanks to a new framework that disentangles true causal effects from spurious correlations by jointly tackling observable and unobservable confounders.

LLMs can find more real-world firmware vulnerabilities (and with higher precision) when structured as a feedback-driven system that interleaves reasoning and tool interaction, rather than a one-pass static analysis.

Current image quality metrics struggle to articulate *why* one high-quality image is better than another, but this challenge shows MLLMs are closing the gap by providing expert-level explanations.