avatar

Hui-Po Wang

Research Scientist
Meta
huipowang24(at)gmail(dot)com


Research Scientist at Meta.

My research focuses on efficient and trustworthy machine learning, aiming to make AI models more accessible and beneficial to the general public. Recently, I have been exploring the safety and reliability of Large Language Models (LLMs), including defenses against prompt injection attacks, protection strategies, and methods to identify, mitigate, and even leverage implicit biases within these systems.

I studied my PhD under the supervision of Prof. Mario Fritz at CISPA, Germany. Before joining CISPA, I was a research intern at the Max-Planck Institute for Informatics, Germany, and received my Master’s degree under the supervision of Prof. Wen-Hsiao Peng and Prof. Wei-Chen Chiu in National Yang Ming Chiao Tung University, Taiwan.

News

Publications [Google Scholar] [DBLP]

  1. ICML
    Stealix: Model Stealing via Prompt Evolution
    Zhixiong Zhuang, Hui-Po Wang, Maria-Irina Nicolae, Fritz Mario
    International Conference on Machine Learning (ICML), 2025
    PDF Code

  2. MICCAI
    Automated Detection of Abnormalities in Zebrafish Development
    Sarath Sivaprasad, Hui-Po Wang, Anna-lisa Jäckel, Jonas Baumann, Carole Baumann, Jennifer Herrmann, Mario Fritz
    Medical Image Computing and Computer Assisted Intervention (MICCAI), 2025

  3. TMLR
    DP-2Stage: Adapting Language Models as Differentially Private Tabular Data Generators
    Tejumade Afonja, Hui-Po Wang, Raouf Kerkouche, Mario Fritz
    Transactions on Machine Learning Research (TMLR), 2025
    PDF Code

  4. NeurIPS
    Language Models as Zero-shot Lossless Gradient Compressors: Towards General Neural Parameter Prior Models
    Hui-Po Wang, Mario Fritz
    Advances in Neural Information Processing Systems (NeurIPS), 2024
    PDF Code

  5. PoPETS
    FedLAP-DP: Federated learning by sharing differentially private loss approximations
    Hui-Po Wang, D Chen, R Kerkouche, M Fritz
    Proceedings on Privacy Enhancing Technologies (PoPETs), 2024
    PDF Code BibTex

  6. ICML
    ProgFed: Effective, Communication, and Computation Efficient Federated Learning by Progressive Training
    Hui-Po Wang, Sebastian U. Stich, Yang He, Mario Fritz
    International Conference on Machine Learning (ICML), 2022
    PDF Code Project Page BibTex
    Spotlight

  7. CVPR
    Hijack-GAN: Unintended-Use of Pretrained, Black-Box GANs
    Hui-Po Wang, Ning Yu, Mario Fritz
    IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021.
    PDF Code Project Page BibTex

  8. CVPR
    Video Rescaling Networks with Joint Optimization Strategies for Downscaling and Upscaling
    Yancheng Huang, Yi-Hsin Chen, Cheng-You Lu, Hui-Po Wang, Wen-Hsiao Peng, Ching-Chun Huang
    IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021.
    PDF Code Project Page BibTex

  9. CVPRw
    InfoScrub: Towards Attribute Privacy by Targeted Obfuscation
    Hui-Po Wang, Tribhuvanesh Orekondy, and Mario Fritz
    IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2019
    PDF BibTex

  10. CVPR
    All about Structure: Adapting Structural Information across Domains for Boosting Semantic Segmentation
    Wei-Lun Chang*, Hui-Po Wang*, Wen-Hsiao Peng, and Wei-Chen Chiu. (*euqal contribution)
    IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019.
    PDF Code Project Page BibTex

  11. APSIPA ASC
    Learning Priors for Adversarial Autoencoders
    Hui-Po Wang, Wen-Hsiao Peng, and Wei-Jan Ko
    APSIPA Annual Summit and Conference (ASC), 2018.
    PDF BibTex
    Best Paper Award Runner-up; Invited to ASIPA T-SIP, 2020


Last Update: