Jiachen's Home Page

Hi, I'm Jiachen Lei 👋

I hold an M.S. degree from Cyberspace Security Lab at Zhejiang University. During my time there, I had the privilege of being mentored by Professor Zhongjie Ba.

My academic journey has been enriched by diverse experiences, including an internship at Microsoft Research under the expert guidance of Shuang Ma, as well as an internship in Professor Anima Anandkumar's group, where I had the privilege of receiving valuable support from Professor Anima Anandkumar, Julius Berner and Professor Chaowei Xiao.

Currently, I am passionate about:

  • Generative Model: Efficiency, Theory and its Application.
  • Multi-modality: Unified Training Framework
  • Security: AIGC Pre/Post-defense, Robustness


Beijing, China
Last updated on Feb 2nd, 2025
Jiachen's profile picture

This is me at ICCV2023

Publications

(ICLR2025) Robust Representation Consistency Model via Contrastive Denoising

  • Jiachen Lei, Julius Berner, Jiongxiao Wang, Zhongzhu Chen, Chaowei Xiao, Zhongjie Ba, Kui Ren, Jun Zhu, Anima Anandkumar
    A unified training framework for representation learning and generative modeling, preliminarily applied to improve model's certified robustness.
    Paper Link: https://arxiv.org/abs/2501.13094

(Neurips2025) Consistency Purification: Effective and Efficient Diffusion Purification towards Certified Robustness

Masked Diffusion Models are Fast Distribution Learners

  • Jiachen Lei, Qinglong Wang, Peng Cheng, Zhongjie Ba*, Zhan Qin, Zhibo Wang, Zhenguang Liu, Kui Ren
    Significantly reducing the training expense of diffusion transformer model by 75%
    Paper Link: https://arxiv.org/abs/2309.14122

(CCS2024) SurrogatePrompt: Bypassing the Safety Filter of Text-To-Image Models via Substitution

Experience

Zhejiang University

Sep 2020 - Mar 2023

Master, Artificial Intelligence

Shanghai Unversity of Electric Power

Sep 2016 - June 2020

Undergraduate, Electrical Engineering