Scholar
Binghui Li
Google Scholar ID: U6BRIM4AAAAJ
CMLR, Peking University
Citations & Impact
Citations
74
H-index
4
i10-index
2
Publications
10
Co-authors
19
list available
Publications
Resume (English only)
Academic Achievements
- (NeurIPS 2025, Spotlight) Functional Scaling Laws in Kernel Regression: Loss Dynamics and Learning Rate Schedules
- (Preprint) M2IO-R1: An Efficient RL-Enhanced Reasoning Framework for Multimodal Retrieval Augmented Multimodal Generation
- (Preprint) New Sphere Packings from the Antipode Construction
- (ICML 2025) On the Clean Generalization and Robust Overfitting in Adversarial Training from Two Theoretical Views: Representation Complexity and Training Dynamics
- (SIGIR 2025) MRAMG-Bench: A Comprehensive Benchmark for Advancing Multimodal Retrieval-Augmented Multimodal Generation
- (ICLR 2025) Feature Averaging: An Implicit Bias of Gradient Descent Leading to Non-Robustness in Neural Networks
- (ICLR 2025) Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data
- (NeurIPS 2022) Why Robust Generalization in Deep Learning is Difficult: Perspective of Expressive Power
- (Preprint) Reconstruction Task Finds Universal Winning Tickets
- (Preprint) Boosting Certified ℓ∞ Robustness with EMA Method and Ensemble Model
Research Experience
- During his Ph.D., he has been involved in multiple research projects and has given several invited talks at academic conferences.
Background
- A third-year Ph.D. student at the Center for Machine Learning Research (CMLR), Peking University, advised by Prof. Liwei Wang and Prof. Lei Wu. Also worked closely with Prof. Yuanzhi Li and Prof. Ruoyu Sun. Research area is machine learning, with special interests in understanding deep learning and algorithms inspired by theoretical insights.
Miscellany
- Contact: libinghui@pku.edu.cn or WeChat
