Hongbin Liu
{first_name}.{last_name}@citadelsecurities.com

Hi! I am a Quantitative Researcher at Citadel Securities. Prior to that, I was a Research Scientist at Google DeepMind.

My research and experience span the life cycle of LLMs - from pre-training and post-training to evaluation. I contributed to Gemini 3 Pro, Gemini 2.5 Pro, Gemini Live, and Personal Intelligence.

I received my Ph.D. from Duke University in 2025, advised by Prof. Neil Gong. Earlier, I earned a B.Eng. from the University of Science and Technology of China in 2020.

2026

• Leave My Images Alone: Preventing Multi-Modal Large Language Models from Analyzing Images via Visual Prompt Injection
  Zedian Shao*, Hongbin Liu*, Yuepeng Hu, and Neil Zhenqiang Gong.
  Annual Meeting of the Association for Computational Linguistics (ACL).

2025

• Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities
  Gemini Team, Google.
  Technical Report.


• Tracing Back the Malicious Clients in Poisoning Attacks to Federated Learning
  Yuqi Jia, Minghong Fang, Hongbin Liu, Jinghuai Zhang, Neil Zhenqiang Gong.
  Conference on Neural Information Processing Systems (NeurIPS).


• Differentially Private Parameter-Efficient Fine-tuning for Large ASR Models
  Hongbin Liu, Lun Wang, Om Thakkar, Abhradeep Thakurta, Arun Narayanan.
  Conference of the International Speech Communication Association (Interspeech).


• Enhancing Prompt Injection Attacks to LLMs via Poisoning Alignment
  Zedian Shao*, Hongbin Liu*, Jaden Mu, Neil Zhenqiang Gong.
  ACM Workshop on Artificial Intelligence and Security (AISec).

2024

• AudioMarkBench: Benchmarking Robustness of Audio Watermarking
  Hongbin Liu*, Moyang Guo*, Zhengyuan Jiang, Lun Wang, Neil Zhenqiang Gong.
  Datasets and Benchmarks Track in Conference on Neural Information Processing Systems (NeurIPS).


• Mudjacking: Patching Backdoor Vulnerabilities in Foundation Models
  Hongbin Liu, Michael K Reiter, Neil Zhenqiang Gong.
  USENIX Security Symposium.


• Visual Hallucinations of Multi-modal Large Language Models
  Wen Huang*, Hongbin Liu*, Minxin Guo, Neil Zhenqiang Gong.
  Findings of the Association for Computational Linguistics (ACL)


• Differentially Private Parameter-Efficient Fine-tuning for Large ASR Models
  Hongbin Liu, Lun Wang, Om Thakkar, Abhradeep Guha Thakurta, and Arun Narayanan.
  DLSP @ IEEE Symposium on Security and Privacy.


• Pre-trained Encoders in Self-Supervised Learning Improve Secure and Privacy-preserving Supervised Learning
  Hongbin Liu*, Wenjie Qu*, Jinyuan Jia, Neil Zhenqiang Gong.
  Security Architectures for Generative Artificial Intelligence (SAGAI) @ IEEE Symposium on Security and Privacy.


• CorruptEncoder: Data Poisoning based Backdoor Attacks to Contrastive Learning
  Jinghuai Zhang, Hongbin Liu, Jinyuan Jia, Neil Zhenqiang Gong.
  IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).


• 10 Security and Privacy Problems in Large Foundation Models
  Jinyuan Jia, Hongbin Liu, Neil Zhenqiang Gong.
  Book Chapter @ AI Embedded Assurance for Cyber Systems.

2023

• Generation-based fuzzing? Don’t build a new generator, reuse!
  Chengbin Pang, Hongbin Liu, Yifan Wang, Neil Zhenqiang Gong, Bing Mao, and Jun Xu.
  Computers & Security


• PointCert: Point Cloud Classification with Deterministic Certified Robustness Guarantees
  Jinghuai Zhang, Jinyuan Jia, Hongbin Liu, and Neil Zhenqiang Gong.
  IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

2022

• PoisonedEncoder: Poisoning the Unlabeled Pre-training Data in Contrastive Learning
  Hongbin Liu, Jinyuan Jia, and Neil Zhenqiang Gong.
  USENIX Security Symposium.


• Semi-Leak: Membership Inference Attacks Against Semi-supervised Learning
  Xinlei He, Hongbin Liu, Neil Zhenqiang Gong, and Yang Zhang.
  European Conference on Computer Vision (ECCV).


• StolenEncoder: Stealing Pre-trained Encoders in Self-supervised Learning
  Yupei Liu, Jinyuan Jia, Hongbin Liu, Neil Zhenqiang Gong.
  ACM Conference on Computer and Communications Security (CCS).


• Almost Tight L0-norm Certified Robustness of Top-k Predictions against Adversarial Perturbations
  Jinyuan Jia, Binghui Wang, Xiaoyu Cao, Hongbin Liu and Neil Zhenqiang Gong.
  International Conference on Learning Representations (ICLR).

2021

• EncoderMI: Membership Inference against Pre-trained Encoders in Contrastive Learning
  Hongbin Liu*, Jinyuan Jia*, Wenjie Qu and Neil Zhenqiang Gong.
  ACM Conference on Computer and Communications Security (CCS)


• PointGuard: Provably Robust 3D Point Cloud Classification
  Hongbin Liu*, Jinyuan Jia* and Neil Zhenqiang Gong.
  IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)


• On the Intrinsic Differential Privacy of Bagging
  Hongbin Liu, Jinyuan Jia and Neil Zhenqiang Gong.
  International Joint Conference on Artificial Intelligence (IJCAI)

2026

• Robustness of Vision Foundation Models to Common Perturbations
  Hongbin Liu, Zhengyuan Jiang, Cheng Hong, Neil Zhenqiang Gong.
  


• Steering LLMs for Culturally Localized Generation
  Simran Khanuja, Hongbin Liu, Shujian Zhang, John Lambert, Mingqing Chen, Rajiv Mathews, Lun Wang.
  

2025

• Automatically Generating Visual Hallucination Test Cases for Multimodal Large Language Models
  Zhongye Liu*, Hongbin Liu*, Yuepeng Hu, Zedian Shao, Neil Zhenqiang Gong.
  


• Can DeepFake Speech be Reliably Detected?
  Hongbin Liu*, Youzheng Chen, Arun Narayanan, Athula Balachandran, Pedro J. Moreno, Lun Wang*.
  

Conference Program Committee: CCS (2025), ICML (2025), ICCV (2025, 2023), ICLR (2025), ECCV (2024), ACCV (2024), AACD (2024).

Journal Program Committee: TOIT (2025), Computer (2025), TIFS (2025), TOPS (2025).