Feng He (何峰)

hefengcs [at] mail.ustc [dot] edu [dot] cn

I received my M.S. in Neuroscience from the University of Science and Technology of China in June 2026. My research focuses on reliable reasoning and agentic behavior in LLMs, especially their behavior under false premises, hallucinated assumptions, and unreliable verification signals.

I was advised by Prof. Quan Wen and worked closely with Dr. Qiankun Li on scientific foundation models and multimodal representations. Prior to USTC, I received my B.E. in Computer Science from Yangtze University in June 2023.

Google Scholar / GitHub / LinkedIn / CV

Research Highlights

Reliable LLM Reasoning

I study how LLMs behave under false premises, hallucinated assumptions, and unreliable verification signals. Manuscripts are available upon request.

Verification Trap: Understanding Test-Time Selection Failures under False Premises in Code Generation
Feng He, Hejia Wang, Linghao Meng, Ming Gao, Qiankun Li

Locked Behavior: Understanding False-Premise Failures in Code Generation
Feng He, Linghao Meng, Junyuan Mao, Kun Wang, Liang Lin, Ruixiang Tang, Qiankun Li

LLM Hallucinations
Feng He et al.

Scientific and Multimodal AI

From Pixels to Views: Learning Angular-Aware and Physics-Consistent Representations for Light Field Microscopy
Feng He, Guodong Tan, Qiankun Li, Jun Yu, Quan Wen
NeurIPS 2025 / pdf / code

BeautyDiffusion: Generative Latent Decomposition for Makeup Transfer via Diffusion Models
Feng He, Hanlin Li, Xin Ning, Qiankun Li
Information Fusion / pdf

Full publication list available on Google Scholar.

Awards

China National Scholarship (Top 0.1% nationwide), 2026

Academic Service

Conference Reviewer: NeurIPS 2026, ICLR 2026

Journal Reviewer: IEEE Transactions on Multimedia, Neural Networks, ACM TOMM, Knowledge-Based Systems, The Visual Computer, and others

Education

University of Science and Technology of China (USTC)
M.S. in Neuroscience, June 2026

Yangtze University
B.E. in Computer Science, June 2023

Last updated: June 2026 | Template credit