My research lies at the intersection of theoretical computer science, discrete mathematics, and formal methods. I am deeply interested in developing rigorous mathematical and algorithmic frameworks to solve complex system problems—ranging from structural graph theory to AI safety.

More information about Yanni Dong

Currently, my work focuses on applying formal methods and model checking to hallucination detection in Large Language Models (LLMs).

Research Pillars
1. Foundational Theory: Graph Algorithms & Combinatorics
• Graph Algorithms & Complexity: Investigating the algorithmic boundaries and structural properties of complex networks.
• Extremal Combinatorics: Analyzing extremal problems with a specific focus on graphical/topological indices.
2. Applied Formal Methods: Trustworthy AI & System Reliability
• LLM Hallucination Detection: Leveraging model checking, verification techniques, and graph theory to mathematically detect and bound factual inconsistencies in generative AI.
• Fault Diagnostics: Utilizing Fault Tree Analysis and formal verification for fault deception, isolation, and recovery (FDIR) in critical systems.