Jiaer Xia

Hi! I'm Jiaer Xia, a second-year PhD student at the Department of Computer Science, Hong Kong Baptist University, supervised by Prof. Kaiyang Zhou.

Previously, I was a Master student at the MACLab at Xiamen University, supervised by Prof. Rongrong Ji. I received my Bachelor's Degree in Engineering (2017-2021) from the Donghua University, working closely with Prof. Mingbo Zhao.

My current research mainly focuses on Multimodal LLMs. Feel free to contact me.

Email (jackson.xje@gmail.com) / Google Scholar / GitHub / Twitter

News

[7/2025] GCoT selected as Highlight Paper at ICCV 2025.
[6/2025] GCoT accepted by ICCV 2025.
[5/2025] Released Visionary-R1, a new RL method for visual reasoning.

Older News & Activities

[3/2024] CycleTrans accepted by TNNLS.
[12/2023] Two papers, ADP and SPT, accepted by AAAI 2024.

Publications

	Bootstrapping Grounded Chain-of-Thought in Multimodal LLMs for Data-Efficient Model Adaptation Jiaer Xia, Bingkui Tong, Yuhang Zang, Rui Shao, Kaiyang Zhou^† ICCV, 2025 (Highlight) Paper / Code / Blog An annotation-free bootstrapping method that enhances your CoT with grounding information, enabling rapid adaptation.

	Visionary-R1: Mitigating Shortcuts in Visual Reasoning with Reinforcement Learning Jiaer Xia, Yuhang Zang, Peng Gao, Yixuan Li, Kaiyang Zhou^† Preprint, 2025 Paper / Code / Blog Pure reinforcement learning for visual reasoning without chain-of-thought annotations.

	Mitigating Hallucination in Multimodal LLMs with Layer Contrastive Decoding Bingkui Tong, Jiaer Xia, Kaiyang Zhou Neurips, 2025, Workshop Paper / Code A training-free decoding approach that uses shallow-layer features as negative references in contrastive decoding to reduce hallucinations.

	Measuring Epistemic Humility in Multimodal Large Language Models Bingkui Tong, Jiaer Xia, Sifeng Shang, Kaiyang Zhou Preprint, 2025 Paper / Code Blog A new benchmark that encompasses a broader spectrum of hallucination types and provides a more realistic assessment of MLLM reliability in safety-critical settings.

	Attention Disturbance and Dual-Path Constraint Network for Occluded Person Re-Identification Jiaer Xia, Lei Tan, Pingyang Dai^†, Mingbo Zhao, Yongjian Wu, Liujuan Cao AAAI, 2024 Paper / Arxiv Attention disturbance and dual-path constraint network for occluded person re-identification.

	Occluded Person Re-Identification via Saliency-Guided Patch Transfer Lei Tan, Jiaer Xia, Wenfeng Liu, Pingyang Dai^†, Yongjian Wu, Liujuan Cao AAAI, 2024 Paper / Arxiv Saliency-guided patch transfer for occluded person re-identification.

	CycleTrans: Learning Neutral yet Discriminative Features for Visible-Infrared Person Re-Identification Qiong Wu, Jiaer Xia, Pingyang Dai^†, Yiyi Zhou, Yongjian Wu, Rongrong Ji TNNLS, 2024 Paper / Arxiv Learning neutral yet discriminative features for VI-ReID.

Activities

Conference Reviewer: CVPR, ECCV, ICCV, NeurIPS, ICLR, ICML, AAAI, AISTATS
Journal Reviewer: TPAMI, IJCV, TCSVT

Teaching

Hong Kong Baptist University:
- COMP7065 (Teaching Assistant)
- COMP2045 & COMP2046 (Teaching Assistant)
- Undergraduate Final Year Project

Last updated in Nov. 2025.

Homepage credits: Jon Barron.