ZHANG XIN 张鑫
👩🏻💻Profile
I am currently a Research Scientist at CFAR, A*STAR. Prior to that, I was a visiting student at A*STAR, working with Prof. Joey Tianyi Zhou and Dr. Jiawei Du.
I received my Ph.D. degree in December 2024 from Xidian University, under the supervision of Prof. Yunsong Li and Prof. Weiying Xie. Before that, I obtained my Bachelor’s degree in 2019, also from Xidian University.
My research interests lie in machine learning and computer vision, with a focus on data-centric efficiency, generative AI, and deepfake generation & detection.
I am also looking for self-motivated PhD students, research interns, and visiting students. In addition, A*STAR offers a variety of scholarships.
⚡️ News
2026.9
📌 Three of our works are accepted by NeurIPS 2026! 🎉 "Visual Latents Know More Than They Say: Unsilencing Latent Reasoning in MLLMs", "Efficient Dataset Distillation for Pre-Trained Self-Supervised Models via Statistical Flow Matching", and "On the Feasibility of Identity Manipulation for Diffusion-Based Face Privacy Preservation". Thanks to all my wonderful collaborators!
2026.9
📌 Our work "Value-Aware KV-Cache Lifecycle Management for Multi-Turn Agentic LLM Serving" is accepted by IEEE HPCC 2026 as a regular full paper (17.67% acceptance rate)! 🎉
2026.9
📌 Our Frontiers in Robotics and AI Research Topic Vision-Language-Action Models and World Models for Robotic Manipulation, co-edited with Dr Moyun Liu, Dr Jiawei Du and Dr Mobarak I. Hoque, is now open for submissions! Deadline: 29 March 2027. Link
2026.6
📌 Our work "Agentic Tool-Augmented Reasoning for Explainable Image Forgery Detection" is accepted by ACM MM 2026 as an oral presentation! 🎉
2026.6
📌 Our work "Mitigate Modality-Asymmetric Forgetting via Stabilizing Visual Representations in CLIP-Based Class-Incremental Learning" is accepted by ECCV 2026! 🎉
2026.6
📌 Our work "Draw2Think: Harnessing Geometry Reasoning through Constraint Engine Interaction" is accepted by the ICML 2026 AI4Math Workshop! 🎉 Paper
2026.6
📌 Our Frontiers Research Topic Towards Trustworthy Visual Foundation Models in the Real World, co-edited with Prof. Joey Tianyi Zhou and Prof. Ismail Ben Ayed, is now open for submissions! Link
2026.5
📌 Our work "Visual Latents Know More Than They Say: Unsilencing Latent Reasoning in MLLMs" is posted online! Paper
2026.5
📌 Our work "Draw2Think: Harnessing Geometry Reasoning through Constraint Engine Interaction" is posted online! Paper
2026.4
📌 Our work "Mitigating Entangled Steering in Large Vision-Language Models for Hallucination Reduction" is posted online! Paper
2026.4
📌 Our book "Weakly Supervised Learning-based Hyperspectral Image Anomaly/Target Detection" is posted online! Book
2025.10
📌 I will be serving as the session chair for Generative AI: Generative Multimedia IV and VI in ACMMM 2025. I look forward to meeting you in Dublin 🇮🇪!
2025.9
2025.8
📌 Our ICLR work “Breaking Class Barriers” was presented at the Workshop on Tensor Representation for Machine Learning (AIP, RIKEN 🇯🇵). Thanks to Prof. Qibin Zhao for the invitation!
2025.5
📌 Our work "Beyond Modality Collapse: Representations Blending for Multimodal Dataset Distillation" is posted online! Paper
2025.3
📌 Honored to receive the 2024 A*STAR Career Development Fund (CDF) award as PI. Sincere thanks to my collaborators and mentors for their continued support and guidance. News
2025.1
2024.12
📌 Successfully defended my dissertation and obtained my Ph.D. degree! 🎓
2024.11
📌 Win National Scholarship for PhD Student again!
2024.10
📌 Our dataset pruning work published on ACMM has been nominated for Best Paper Award!
2024.9
📌 Our work "Diversity-Driven Synthesis: Enhancing Dataset Distillation through Directed Weight Adjustment" is accepted by NeurIPS (spotlight)! Paper
2024.8
📌 Our work "Breaking Class Barriers: Efficient Dataset Distillation via Inter-Class Feature Compensator" is posted online! Paper
2024.7
📌 Our work "Markov-PQ: Joint Pruning-Quantization via Learnable Markov Chain" is accepted by IEEE TCSVT! Paper
2024.7
📌 Our work "Evolution-aware VAriance (EVA) Coreset Selection for Medical Image Classification" is accepted by ACMM (oral, best paper nomination)! Paper
2024.2
2023.10
📌 Win National Scholarship for PhD Student!
📝Publications
For the full list, please refer to my Google Scholar page: Google Scholar
Agentic Systems & Reasoning
[IEEE HPCC 2026] Value-Aware KV-Cache Lifecycle Management for Multi-Turn Agentic LLM Serving.
Xin Zhang, Jiawei Du, Joey Tianyi Zhou.
Xin Zhang, Jiawei Du, Joey Tianyi Zhou.
[ACM MM 2026 (Oral)] Agentic Tool-Augmented Reasoning for Explainable Image Forgery Detection.
Zhiya Tan, Jing Huang, Changtao Miao, Lin Tan, Xin Zhang, Weiwei Feng, Jianshu Li, Joey Tianyi Zhou.
Zhiya Tan, Jing Huang, Changtao Miao, Lin Tan, Xin Zhang, Weiwei Feng, Jianshu Li, Joey Tianyi Zhou.
[ICML 2026 AI4Math Workshop] Draw2Think: Harnessing Geometry Reasoning through Constraint Engine Interaction.
Juncheng Hu, Jiawei Du, Xin Zhang, Joey Tianyi Zhou. Paper
Juncheng Hu, Jiawei Du, Xin Zhang, Joey Tianyi Zhou. Paper
Vision-Language & Multimodal Learning
[ECCV 2026] Mitigate Modality-Asymmetric Forgetting via Stabilizing Visual Representations in CLIP-Based Class-Incremental Learning.
Yuanhong Zhang, Yanan Chen, Xin Zhang, Zhaoyang Wang, Weizhan Zhang, Muyao Yuan, Hongjin Niu, Lan Ma, Yuan Gao, Joey Tianyi Zhou.
Yuanhong Zhang, Yanan Chen, Xin Zhang, Zhaoyang Wang, Weizhan Zhang, Muyao Yuan, Hongjin Niu, Lan Ma, Yuan Gao, Joey Tianyi Zhou.
[NeurIPS 2026] Visual Latents Know More Than They Say: Unsilencing Latent Reasoning in MLLMs.
Xin Zhang, Qiqi Tao, Jiawei Du, Moyun Liu, Joey Tianyi Zhou. Paper
Xin Zhang, Qiqi Tao, Jiawei Du, Moyun Liu, Joey Tianyi Zhou. Paper
Data-centric Efficiency
[NeurIPS 2026] Efficient Dataset Distillation for Pre-Trained Self-Supervised Models via Statistical Flow Matching.
Qianxin Xia, Jiawei Du, Yuhan Zhang, Xin Zhang, Xuewan He, Wenbo Jiang, Jielei Wang, Tao Luo, Guoming Lu.
Qianxin Xia, Jiawei Du, Yuhan Zhang, Xin Zhang, Xuewan He, Wenbo Jiang, Jielei Wang, Tao Luo, Guoming Lu.
[NeurIPS 2025] Beyond Modality Collapse: Representations Blending for Multimodal Dataset Distillation.
Xin Zhang, Ziruo Zhang, Jiawei Du, Zuozhu Liu, Joey Tianyi Zhou.
Xin Zhang, Ziruo Zhang, Jiawei Du, Zuozhu Liu, Joey Tianyi Zhou.
[ICLR 2025] Breaking Class Barriers: Efficient Dataset Distillation via Inter-class Feature Compensator.
Xin Zhang, Jiawei Du, Ping Liu, Joey Tianyi Zhou.
Xin Zhang, Jiawei Du, Ping Liu, Joey Tianyi Zhou.
[NeurIPS 2024 Spotlight] Diversity-Driven Synthesis: Enhancing Dataset Distillation through Directed Weight Adjustment.
Jiawei Du, Xin Zhang, Juncheng Hu, Wenxin Huang, Joey Tianyi Zhou.
Jiawei Du, Xin Zhang, Juncheng Hu, Wenxin Huang, Joey Tianyi Zhou.
[ACM MM 2024 (Oral)] Evolution-aware VAriance (EVA) Coreset Selection for Medical Image Classification.
Yuxin Hong, Xiao Zhang, Xin Zhang, Joey Tianyi Zhou.
Yuxin Hong, Xiao Zhang, Xin Zhang, Joey Tianyi Zhou.
[CVPR 2024] Spanning Training Progress: Temporal Dual-Depth Scoring (TDDS) for Enhanced Dataset Pruning.
Xin Zhang, Jiawei Du, Yunsong Li, Weiying Xie, Joey Tianyi Zhou.
Xin Zhang, Jiawei Du, Yunsong Li, Weiying Xie, Joey Tianyi Zhou.
Generative AI & Media Forensics
[NeurIPS 2026] On the Feasibility of Identity Manipulation for Diffusion-Based Face Privacy Preservation.
Xuemei Jia, Jiawei Du, Jiawei Liu, Xin Zhang, Jun Chen, Zheng Wang, Joey Tianyi Zhou.
Xuemei Jia, Jiawei Du, Jiawei Liu, Xin Zhang, Jun Chen, Zheng Wang, Joey Tianyi Zhou.
[ECCV 2026 Workshop] Fine-Grained Bias Manipulation of Text-to-Image Models via Counterfactual Data Synthesis.
Takuto Onikubo, Xin Zhang, Joey Tianyi Zhou, Shin'ichi Satoh.
Takuto Onikubo, Xin Zhang, Joey Tianyi Zhou, Shin'ichi Satoh.
Model-centric Efficiency
[IEEE TNNLS 2023] Block-Wise Partner Learning for Model Compression.
Xin Zhang, Weiying Xie, Yunsong Li, Jie Lei, Kai Jiang, Leyuan Fang, Qian Du.
Xin Zhang, Weiying Xie, Yunsong Li, Jie Lei, Kai Jiang, Leyuan Fang, Qian Du.
[IEEE TIP 2023] Reaf: Remembering enhancement and entropy-based asymptotic forgetting for filter pruning.
Xin Zhang, Weiying Xie, Yunsong Li, Kai Jiang, Leyuan Fang.
Xin Zhang, Weiying Xie, Yunsong Li, Kai Jiang, Leyuan Fang.
[IEEE TGRS 2023] Co-compression via superior gene for remote sensing scene classification.
Weiying Xie, Xiaoyi Fan, Xin Zhang, Yunsong Li, Min Sheng, Leyuan Fang.
Weiying Xie, Xiaoyi Fan, Xin Zhang, Yunsong Li, Min Sheng, Leyuan Fang.
[IEEE TC 2021] Filter pruning via learned representation median in the frequency domain.
Xin Zhang, Weiying Xie, Yunsong Li, Jie Lei, Qian Du.
Xin Zhang, Weiying Xie, Yunsong Li, Jie Lei, Qian Du.
Preprints
[Preprint] SkillGLoW: Procedural-Family Skill Consolidation for Self-Improving Agents on Long-Horizon Task Streams.
Ao Yan, Xin Zhang, Jiawei Du, Joey Tianyi Zhou. Paper
Ao Yan, Xin Zhang, Jiawei Du, Joey Tianyi Zhou. Paper
[Preprint] Mitigating Entangled Steering in Large Vision-Language Models for Hallucination Reduction.
Yuanhong Zhang, Zhaoyang Wang, Xin Zhang, Weizhan Zhang, Sha Gong, Yuan Gao, Joey Tianyi Zhou. Paper
Yuanhong Zhang, Zhaoyang Wang, Xin Zhang, Weizhan Zhang, Sha Gong, Yuan Gao, Joey Tianyi Zhou. Paper
Image Processing
[IEEE TC 2021] Weakly supervised low-rank representation for hyperspectral anomaly detection.
Weiying Xie, Xin Zhang, Yunsong Li, Jie Lei, Jiaojiao Li, Qian Du.
Weiying Xie, Xin Zhang, Yunsong Li, Jie Lei, Jiaojiao Li, Qian Du.
[IEEE Jstar 2020] Background learning based on target suppression constraint for hyperspectral target detection.
Weiying Xie, Xin Zhang, Yunsong Li, Keyan Wang, Qian Du.
Weiying Xie, Xin Zhang, Yunsong Li, Keyan Wang, Qian Du.