• I am a research scientist and tech lead at ByteDance (Seed), specializing in enhancing the alignment of large language models (LLMs), with a particular focus on cross-modal alignment using reinforcement learning. My work encompasses Reinforcement Learning from Human Feedback (RLHF) for multimodal LLMs, as well as exploring the vulnerabilities and safety challenges within current LLM systems. Additionally, I am passionate about developing innovative applications for LLM-based agents.

• I earned my PhD from Nanyang Technological University (NTU, Singapore), where my research centered on “Adversarial Robustness of Deep Reinforcement Learning.”

Research Interests: Multimodal LLM, Reinforcement Learning, Trustworthy AI

Recent News

• One paper accepted by KDD-2025. Paper Title: Stabilizing Modality Gap & Lowering Gradient Norms Improves Zero-Shot Adversarial Robustness of VLMs.
• Our paper has been nominated as the best paper in SIGIR-2024. Title: Adaptive In-Context Learning with Large Language Models for Bundle Generation.
• [June 2024] I return Bytedance as a research scientist and tech lead, working on the RLHF alignment of multimodal LLM.
• Two papers accepted by SIGIR-2024.
• My research work on audio watermarking was posted on Linkedin
• One paper accepted by NeurIPS-2023 with the title “Unsupervised Video Domain Adaptation for Action Recognition: A Disentanglement Perspective”.
• I was invited by AI-TIME to give a talk on 31 Aug 2023.
• Invited as reviewer for ICLR-2024, ICML-2024, KDD-2024, NeurIPS-2024.
• Our work on ChatGPT driven Voice base Conversational Recommender Systems has been posted by several Chinese media platforms, such as 语音之家, 火山语音
• Our work about Audio QR Code has been highlighted by IJCAI-2023 officially, which can be referred on different media platforms. such as 🔬Linked-in, Facebook, Twitter.
• One paper accepted by InterSpeech-2023 with the title “S2CD-VC: Self-heuristic Speaker Content Disentanglement for Any-to-Any Voice Conversion”
• One paper accepted by IJCAI-2023 with the title “AudioQR: Deep Neural Audio Watermarks For QR Code” (acceptance rate 20%).
• One paper accepted by SIGIR-2023 with the title “Towards Building Voice-based Conversational RecommenderSystems: Datasets, Potential Solutions, and Prospects”.
• One paper entitled “DaisyRec 2.0: Benchmarking Recommendation for Rigorous Evaluation” has been accepted by IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI, IF=24.31).
• Invited as program committee member of KDD-2023, NeurIPS-2023, ICML-2023.
• One paper entitled “Dynamic Transfer Gaussian Process Regression” has been accepted as a full paper by [CIKM-2022].
• My research papers are reported by several Chinese media platforms, such as PaperWeekly, 语音之家，深科技，火山引擎.
• One paper entitled “Transfer Kernel Learning for Multi-source Transfer Gaussian Process Regression” has been accepted by IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI, IF=24.31).
• One paper (First Author) entitled “Synthesising Audio Adversarial Examples for Automatic Speech Recognition” has been accepted as full paper by [SIGKDD-2022] (15% of acceptance out of 1695 submissions)
• One paper (First Author) entitled “Importance Prioritized Policy Distillation” has been accepted as full paper by [SIGKDD-2022] (15% of acceptance out of 1695 submissions)
• One paper entitled “Next Point-of-Interest Recommendation with Inferring Multi-step Future Preferences” has been accepted by [IJCAI-2022] (15% of acceptance out of 4535 submissions)
• One paper entitled “Revisiting Bundle Recommendation: Datasets, Tasks, Challenges and Opportunities for Intent-aware Product Bundling” has been accepted by [SIGIR-2022].
• Program committee member of NeurIPS-2022, ICML-2022, SIGKDD-2022.
• One paper accepted by IEEE Transactions on Cybernetics (IF=19.12).
• One paper entitled “Language Adaptive Cross-lingual Speech Representation Learning with Sparse Sharing Sub-networks” has been accepted by [ICASSP-2022].
• One paper entitled “Spiking Pitch Black: Poisoning an Unknown Environment to Attack Unknown Reinforcement Learners” has been accepted by [AAMAS-2022].
• I finished my PhD oral defense in NTU.
• One paper entitled “Adversary Agnostic Robust Deep Reinforcement Learning” has been accepted by [IEEE Transactions on Neural Networks and Learning Systems] (IF=14.26).