Yihang Chen (陈奕行)

I received the M.S. degree from the School of Computer and Communication Sciences at EPFL in February 2024, advised by Prof. Volkan Cevher; and the B.S. degree from the School of Mathematical Sciences at Peking University in July 2021, advised by Prof. Liwei Wang.

I am currently working as an internship student at Laboratory for Information and Inference Systems (LIONS), EPFL.

Email  /  CV  /  Google Scholar  /  Twitter  /  Github  /  LinkedIn

Research

I have a broad interest in deep learning theory and generative models. At present, my research mainly focus on:

• Safety of LLMs: Jailbreaking and Membership Inference Attacks.
• Applied Probability: Stochastic Control, Optimal Transport, Generative Flow Networks.
• AutoML: Pruning, Neural Architecture Search.
• Deep Learning Theory: Overparameterized Neural Networks, Robust Machine Learning.

Publications

* indicates equal contribution

We propose Order-Preserving GFlowNets (OP-GFNs), which sample composite objects given only the (partial) order, instead of the reward function.

We provide the upper bound of the parameter distribution moving and generalization error on the 0-1 classification task of the infinitely deep and wide ResNets.

We sanity-check prune-at-init methods, and find them hardly exploits any information from the training data. We also propose "zero-shot" pruning, which only relies on simple data-independent pruning ratios for each layer.

Miscellanea

Invited Talks

Honors and Awards

Last Modified: March 22th, 2024. Website template from Jon Barron.