Hello~ I am Xinyi, a Ph.D. candidate at University College London (UCL), luckily supervised by Prof. Jing-Hao Xue. Before that, I received my master’s degree with honors from Xiamen University (XMU), fortunately supervised by Dr. Yang Lu.

You can reach me at xinyi.shang.23@ucl.ac.uk, or find my publications on Google Scholar .

🔥 News

  • 2026.04:  📄 We release a technical report for A Systematic and Comprehensive Analysis of Claude Code. Paper at: [Link]
  • 2026.02:  🎉 Two papers are accepted by CVPR 2026.
  • 2025.09:  Started as a visiting student at MBZUAI, supervised by Prof. Zhiqiang Shen.
  • 2023.05:  🎉 Received the Outstanding Master Thesis Award.
  • 2023.03:  🎉 Received the PhD offer from University College London (UCL), supported by full scholarship!
  • 2022.11:  Started a research internship at Westlake University, supervised by Prof. Tao Lin.
  • 2022.09:  Awarded the China National Scholarship.
  • 2022.04:  One IJCAI paper is accepted.
  • 2022.03:  One ICME paper is accepted as oral.

🎯 Research Interests

My research centers on generalized and efficient deep learning. Numbers in brackets link to the corresponding entries in Publications below.

  • Deep Learning Under Imperfect Data — Federated Learning (FL). A privacy-preserving distributed framework that lets multiple devices or organizations collaboratively train a global model without sharing raw data.
    • Imperfect global data scenarios: tackling long-tailed class distributions [1,2] and limited labeled data [5,6] to improve robustness and reliability.
    • Personalization: enhancing local-model personalization while preserving global performance [3].
    • Generalization & optimization: improving global-model generalization through the lens of training dynamics [4].
  • Efficient Deep Learning — Dataset Distillation. Compressing a large dataset into a much smaller distilled one while preserving downstream accuracy.
    • Efficient distillation: methods that scale from CIFAR-10 to ImageNet-1K [8] and privacy-preserving variants [11].
    • Enhanced utilization: unlocking the full label potential of distilled data [7], revealing key advantages beyond training efficiency [9], and novel usage paradigms [10].
    • Survey: the first comprehensive, stage-wise review of dataset distillation in the large-scale-data era [12].
  • Efficient Deep Learning — Data-Efficient Learning. A data-centric perspective on which data improves training efficiency, and how to acquire it.
    • Understanding efficient data: analyzing how samples (X) and targets (Y) jointly shape training efficiency [13].
    • Data optimization: leveraging publicly pre-trained models to optimize large-scale unlabeled data [14].

📝 Publications

* denotes equal contribution; denotes corresponding author.

IJCAI 2022
creff

[1] Federated Learning on Heterogeneous and Long-Tailed Data via Classifier Re-Training with Federated Features | [code]

Xinyi Shang, Yang Lu, Gang Huang, Hanzi Wang.

  • We first find that the biased classifier is the primary factor behind the poor performance of the global model, then propose CReFF to optimize a small set of learnable features for classifier re-training.
ICME 2022 Oral
fedic

[2] FEDIC: Federated Learning on Non-IID and Long-Tailed Data via Calibrated Distillation | [code]

Xinyi Shang, Yang Lu, Yiu-ming Cheung, Hanzi Wang.

  • A new distillation method with logit adjustment and calibration gating network to solve the joint problem of heterogeneous and long-tailed data.
ICCV 2023
fedetf

[3] No Fear of Classifier Biases: Neural Collapse Inspired Federated Learning with Synthetic and Fixed Classifier | [code]

Zexi Li, Xinyi Shang, Rui He, Tao Lin, Chao Wu.

  • A neural-collapse-inspired method that mitigates classifier biases in federated learning, achieving high global-model generalization together with strong local-model personalization.
ICML 2023
fedlaw

[4] Understanding the Training Dynamics in Federated Deep Learning via Aggregation Weight Optimization | [code]

Zexi Li, Tao Lin, Xinyi Shang, Chao Wu.

  • We analyze FL training dynamics through client coherence and global weight shrinking, and design an aggregation algorithm that measurably improves generalization.
CVPR 2025
mindthegap

[5] Mind the Gap: Confidence Discrepancy Can Guide Federated Semi-Supervised Learning Across Pseudo-Mismatch

Yijie Liu, Xinyi Shang, Yiqun Zhang, Yang Lu, Chen Gong, Jing-Hao Xue, Hanzi Wang.

  • We show that (1) data heterogeneity intensifies pseudo-label mismatches, and (2) local- and global-model predictive tendencies diverge with heterogeneity. We propose a simple yet effective method to correct pseudo-labels by exploiting confidence discrepancies.
TAI 2025
fedssl-ah

[6] Federated Semi-Supervised Learning with Annotation Heterogeneity

Xinyi Shang, Gang Huang, Yang Lu, Jian Lou, Bo Han, Yiu-ming Cheung, Hanzi Wang.

  • We formalize Federated Semi-Supervised Learning with annotation heterogeneity and propose a new framework with a mutual-learning strategy.
ICLR 2025
gift

[7] GIFT: Unlocking Full Potential of Labels in Distilled Dataset at Near-zero Cost | [code]

Xinyi Shang*, Peng Sun*, Tao Lin.

  • Models trained on distilled datasets are highly sensitive to the soft-label loss. Building on this insight, we introduce a plug-and-play approach that efficiently leverages full label information at near-zero cost.
ICLRW 2024 DMLR
infocomp

[8] Information Compensation: A Fix for Any-scale Dataset Distillation

Peng Sun, Bei Shi, Xinyi Shang, Tao Lin.

  • A near-lossless information-compression approach that distills the key information of original datasets with minimal loss, surpassing existing methods in both efficiency and effectiveness across dataset scales.
Preprint
earlystage

[9] Early-Stage Training with Distilled Data Helps Prevalently

Xinyi Shang, Jing-Hao Xue.

  • We uncover a key advantage of distilled datasets: in early training phases, models trained on distilled data show substantially higher efficiency than those trained on original data. We provide explanations and validate the practical implications.
Preprint
rethinking

[10] Rethinking the Use of Distilled Data: Embrace, Don’t Abandon, the Original Data

Xinyi Shang, Jing-Hao Xue.

  • We revisit the use of distilled data and propose a dynamic framework that integrates distilled and original data by analyzing their distinct properties. Efficacy is demonstrated both theoretically and empirically.
Preprint
privacy

[11] Privacy as a Free Lunch: Crafting Initial Distilled Datasets through the Kaleidoscope

Shuo Shi*, Peng Sun*, Xinyi Shang*, Tianyu Du, Xuhong Zhang, Jianwei Yin, Tao Lin.

  • We identify explicit privacy leakage in distilled datasets and show theoretically that it stems from initializing distilled images with real data. A plug-and-play module applies strong perturbations to real data during initialization to address this.
Preprint T-PAMI
dd-survey

[12] Dataset Distillation in the Era of Large-Scale Data: Methods, Analysis, and Future Directions

Xinyi Shang, Peng Sun, Zhiqiang Shen, Tao Lin, Jing-Hao Xue.

  • We identify four significant shifts in the field of dataset distillation and provide the first comprehensive, stage-wise review through the dataset-distillation pipeline.
Preprint
equallycritical

[13] Equally Critical: Samples, Targets, and Their Mappings in Datasets

Runkang Yang*, Peng Sun*, Xinyi Shang*, Yi Tang, Tao Lin.

  • Recent work mostly optimizes samples X while overlooking targets Y. We review the sample-target relationship and comprehensively analyze how variations in target and sample types, quantities, and qualities influence training efficiency and efficacy.
Preprint
cudo

[14] Collaborative Unlabeled Data Optimization

Xinyi Shang*, Peng Sun*, Fengyuan Liu*, Tao Lin.

  • We pioneer a data-centric paradigm for collaborative unlabeled data optimization, demonstrating effectiveness and efficiency across diverse datasets and architectures.

🎖 Honors and Awards

  • 2023 Xiamen University Outstanding Master Thesis.
  • 2023 Xiamen University Outstanding Graduates.
  • 2022 China National Scholarship (Top 0.2%, the highest-level scholarship established by the central government).
  • 2022 Excellent Merit Student of Xiamen University (Top 2%).
  • 2021 Merit Student of Xiamen University (Top 8%).
  • 2020 China College Students Innovation and Entrepreneurship Competition — two provincial projects.
  • 2019 Provincial Excellent Volunteer honor (500+ hours of volunteering).
  • 2018 Star of Excellent Volunteers honor (only one student in the college per year).

💻 Research Experience

  • 2022.11 - 2023.09, Research Intern, LINs Lab, Westlake University, supervised by Prof. Tao Lin. Decentralized deep learning.
  • 2022.06 - 2022.09, Research Intern, MARS Lab, Wuhan University, supervised by Prof. Mang Ye. Federated learning.

🤝 Academic Service

  • Conference Reviewer: NeurIPS 2025, ICCV 2025, ICLR 2025, CVPR 2025, WACV 2025, IJCAI 2024.
  • Journal Reviewer: IEEE TNNLS, IEEE TCSVT, ACM CSUR, IEEE TC, IEEE TETCI.

🙌 Voluntary Activities

  • 2019.03 - 2019.09, Director of Teach For China at Zhongnan University of Economics and Law.
  • 2017.09 - 2019.06, Director of We-Bright, supporting 53 rural primary schools across Sichuan and Guangxi provinces.

🎨 Hobbies