Wenxuan Zhang | PhD Student

About Me

I am a forth-year Ph.D. candidate at KAUST in Saudi Arabia. I am part of Vision-CAIR group led by Pr. Mohamed Elhoseiny. My research lies in enhancing the foundation models with emerging abilities and properties, including but not limited to:

Fine-tuning and aligning foundation models with reinforcement learning.
Efficient Training and Continual Learning. Parameter-efficient methods, mitigating forgetting, and learning under constrained computation.
Agents. Building autonomous LLM/VLM agents with enhanced reasoning and tool-use capabilities.

Education

PhD Candidate in Computer Science, King Abdullah University of Science and Technology (KAUST)

MA in Applied Mathematics and Computational Science, 2021, University of Pennsylvania

BSc in Mathematics and Applied Mathematics, 2019, Beijing Normal University

News

Selected Publications [Google Scholar]

Coming Soon

Hybrid Advantage Estimation with Unified Critic for VLM Agentic Reinforcement Learning

Wenxuan Zhang, Yuhui Wang, Donggang Jia, Xiaoqian Shen, Jian Ding, Mohamed Elhoseiny

Coming Soon

dTRPO: Trajectory Reduction in Policy Optimization of Diffusion Large Language Models

Wenxuan Zhang, Lemeng Wu, Changsheng Zhao, Ernie Chang, Mingchen Zhuge, Zechun Liu, DiJia Su, Hanxian Huang, Jun Chen, Chong Zhou, Raghuraman Krishnamoorthi, Vikas Chandra, Mohamed Elhoseiny, Wei Wen

ICLR Spotlight

Bi-Factorial Preference Optimization: Balancing Safety-Helpfulness in Language Models

Wenxuan Zhang, Philip Torr, Mohamed Elhoseiny, Adel Bibi

PDF Code

CVPR

Overcoming Generic Knowledge Loss with Selective Parameter Update

Wenxuan Zhang, Paul Janson, Rahaf Aljundi, Mohamed Elhoseiny

PDF Code Project Page

NeurIPs

Label Delay in Continual Learning

Botos Csaba*, Wenxuan Zhang*, Matthias Müller, Ser-Nam Lim, Mohamed Elhoseiny, Philip Torr, Adel Bibi

PDF Project Page

ICLR

Continual Learning on a Diet: Learning from Sparsely Labeled Streams Under Constrained Computation

Wenxuan Zhang, Youssef Mohamed, Bernard Ghanem, Philip Torr, Adel Bibi, Mohamed Elhoseiny

PDF Code

Services

Reviewers

ICLR, NeurIPs, CVPR, ICCV, TPMAI, ICML
CLAI Unconf, WECIA, ICCVCLVL

Teaching

Teaching Assistant of CS 326, Low Resource Deep Learning, KAUST

Contact

Address: 4335 WS - 11, Building 1 - The Spine, KAUST 23955, Saudi Arabia
Email: wenxuan.zhang@kaust.edu.sa

Last Update:

Adapted from the Minimal Light theme by Yaoyao Liu.