Haicheng Huang 黄海城

I am a rising fourth-year undergraduate student at Shanghai Jiao Tong University, majoring in Information Engineering. I'm currently a research assistant at Shanghai Jiao Tong University, working with Prof. Jiangchao Yao. I was also fortunate to work with Prof. Mingyuan Zhou from UT Austin and mentor Huangjie Zhengduring my undergraduate.

Email  /  Github  /  LinkedIn

Research Interests

My research interests lie in generative models, especially diffusion models and representaion learning. I want to explore:

(1) How can generative models, especially diffusion models and autoregressive models, achieve better and faster performance in various downstream tasks
(2) How to use the synthetic data produced by the generative model or the learned representations to assist in the process of representation learning

I am currently seeking a Ph.D. opportunity for Fall 2026 in computer science!

News

[Sept. 2025] Wino is accepted by NeurIPS 2025 Workshop.
[Sept. 2025] DAR-GDA is accepted by NeurIPS 2025.
[May. 2025] Start internship atCMIC at Shanghai Jiao Tong University .
[July. 2024] Start remote internship at UT Austin.

Publications

We propose DAR-GDA, a three-stage augmentation pipeline that unites model **D**istillation, **A**dversarial alignment, and importance **R**eweighting that makes diffusion-quality augmentation both fast and optimized for improving downstream learning outcomes. Our approach not only surpasses conventional non-foundation-model GDA baselines but also remarkably matches or exceeds the GDA performance of large, web-pretrained text-to-image models, despite using solely in-domain data.

we introduce Wide-In, Narrow-Out (WINO), a training-free decoding algorithm that enables revokable decoding in DLLMs. WINO employs a parallel draft-and-verify mechanism, aggressively drafting multiple tokens while simultaneously using the model's bidirectional context to verify and re-mask suspicious ones for refinement.

The website is built upon this template.