Shuyang Yu

I am Shuyang Yu (於舒扬), a Machine Learning Scientist at Samsung Electronics America. I received my Ph.D. in Computer Science from Michigan State University in 2025, where I was advised by Prof. Jiayu Zhou (currently at the University of Michigan).

My research focuses on building robust and trustworthy AI systems that remain reliable under real-world challenges such as distribution shift, data corruption, data scarcity, and adversarial threats. I am especially interested in methods that help modern AI systems move beyond benchmark performance toward reliable behavior in long-tail, noisy, and high-stakes settings.

Some of my current and recent research directions include:

• Reliable LLM knowledge integration: improving retrieval-augmented reasoning and in-context learning when task-relevant evidence is missing, noisy, or long-tailed, including adaptive information-seeking, robust candidate-set construction, and uncertainty-aware retrieval ranking for knowledge-intensive QA.
• Robust learning across domains: developing methods that help machine learning models generalize under distribution shifts, corrupted data, and limited supervision.
• Trustworthy knowledge transfer and model protection: studying how knowledge moves across models and systems, with an emphasis on robustness, security, and intellectual property protection.

I have collaborated with researchers across data mining, natural language processing, computer vision, healthcare, robotics, and control systems. My work has appeared in venues including ICLR, ICML, KDD, NAACL, ICDM, and IEEE Transactions on Neural Networks and Learning Systems (TNNLS).