Xuefeng Jiang (蒋雪枫)

Xuefeng Jiang

I am a forth-year PhD candidate at Institute of Computing Technology, Chinese Academy of Sciences in Beijing, China. Before that, I received my Bachelor degree from Beijing University of Posts and Telecommunications in the summer of 2021.

I am advised by Prof. Min Liu, associate Prof. Yuwei Wang and associate Prof. Sheng Sun, in the Mobile Computing Team. My research interest mainly includes distributed ML system, data quality, cyber security, AutoDrive and various deep learning applications. I am always open to discussions, please feel free to contact me!

Email / Google Scholar / Github / OpenReview / DBLP / Instagram

Research

Distributed Optimization: Federated Learning, AI Robustness&Fairness
Cyber Security: Code Vulnerability Detection, Malware Detection, Anomaly Detection
Data Quality: Weakly Supervised Learning (Noisy Label Learning & Long-tailed Learning & Semi-supervised Learning & Self-supervised Learning)
Autonomous Driving: Multi-modal 3D Perception, Vision-language Model & LLM for AutoDrive, Multi-sensor Fusion SLAM, Auto Parking, Model Deployment, Dataset Curation, Robotics & Agents
Applications: Recognition and Generation

Services

Conference Program Committee for: ICML, AISTATS, CVPR, TheWebConf (WWW), ICLR, NeurIPS, AAAI, ECAI, ICME, KDD and AAAI workshop
Invited Journal Reviewer for: ACM/TKDD, IEEE/TKDE, IEEE/TNNLS, IEEE/TIV (Transactions on Intelligent Vehicles), ACM Transactions on Modeling and Performance Evaluation of Computing Systems
Help to Review: DCN

Milestones

[2024.11] I was fortunately awarded the National Scholarship 2024, thanks to the guidance of Prof.Liu, associate Prof.Wang and associate Prof.Sun. I also learned a lot from other Ph.D. student candidates.
[2024.10] I visited the Los Angeles and Boise City in the USA. It was a nice chance to go abroad to attend ACM CIKM conference. Many thanks to Prof.Liu!
[2024.09] Two works focusing on long-tailed distributed learning got accepted in TMC and ACML 2024 (~26% acceptance rate).
[2024.08] I checkout from AMD, and focus on research projects in my lab. Many thanks to AMD's colleagues ( especially my mentor Treemann & my colleague Fangyuan & manager Peng ) who once guided and inspired me.
[2024.07] Another work FedELC focusing on noisy clients in federated learning got accepted in ACM/CIKM 2024 full paper track. We also release our benchmark study paper FNBench.
[2024.06] I'm blessed with another chance to breathe again after an accident. Many thanks to my love Iris & my mom & doctor & supervisors & careful friends.
[2024.05] We reach the top 5 (152 participating teams in total) in the CVPR AutoDrive Grand Challenge 2024's Driving with Language Track with Offline labels on the HuggingFace leaderboard.
[2023.11] I finish the first work (about SLAM) explored in AMD with my mentor Treemann&Fangyuan. Though it got rejected by IROS, I still have confidence and regard it as the best project I once enjoyed.
[2023.09] The fairytale unfolds ^_^
[2023.07] It's a very special&meaningful day for me because I meet two idols, and I'll never forget that day ^_^
[2023.06] One co-authored paper FedICT got accepted by IEEE TPDS.
[2023.04] One co-authored paper with HUST & HKUST got accepted by IJCAI 2023.
[2023.01] Two co-authored papers got accepted by IEEE/IPDPS 2023.
[2022.10] I start working at AMD (Advanced Micro Devices, Inc.) AI Group as a Co-op (Cooperator from University) under the guidance of Treemann, in Beijing.
[2022.08] My first work got accepted by ACM/CIKM 2022 with NSF/SIGWeb student award.

Publications & Competitions

More works are on the way : )

Federated Classification Tasks in Long-tailed Data Environments via Classifier Representation Adjustment and Calibration

Xujing Li, Sheng Sun, Min Liu, Ju Ren, Xuefeng Jiang, Tianliu He
IEEE Transactions on Mobile Computing (TMC)
link

To tackle long-tailed data issue in the context of Federated Learning.

FedLF: Adaptive Logit Adjustment and Feature Optimization in Federated Long-Tailed Learning

Xiuhua Lu, Peng Li, Xuefeng Jiang (serves as a corresponding author)
ACML 2024 (~26% acceptance rate)
link

Tackling distributed long-tailed data.

Tackling Noisy Clients in Federated Learning with End-to-end Label Correction

Xuefeng Jiang, Sheng Sun, Jia Li, Jingjing Xue, Runhan Li, Zhiyuan Wu, Gang Xu, Yuwei Wang and Min Liu
ACM CIKM 2024
arXiv

To refine the noisy labels of noisy clients via an end-to-end Optimization framework in the scenario of distributed learning.

AMD's Solution for Driving with Language Challenge

Xuefeng Jiang, Fangyuan Wang, Rongzhang Zheng, Jinzhang Peng, Lu Tian, Emad Barsoum
CVPR 2024 Autonomous Grand Challenge Driving with Language Track. We reach the top 5 teams (152 participating teams in total) with offline labels on HuggingFace leaderboard. link

Multi-modal LLM for AutoDrive Scenarios

Agglomerative Federated Learning: Empowering Larger Model Training via End-Edge-Cloud Collaboration

Zhiyuan Wu, Sheng Sun, Yuwei Wang, Min Liu, Bo Gao, Quyang Pan, Tianliu He, Xuefeng Jiang
IEEE INFOCOM, 2024
link

To utilize cloud-edge-client compuatation to faciliate training larger model agglomeratively.

FedNoRo: Towards Noise-Robust Federated Learning by Addressing Class Imbalance and Label Noise Heterogeneity

Nannan Wu, Li Yu, Xuefeng Jiang, Kwang-Ting Cheng, Zengqiang Yan
International Joint Conference on Artificial Intelligence (IJCAI), 2023
arXiv

To tackle both long-tailed & noisy data issue in the context of Federated Learning.

FedCache: A Knowledge Cache-driven Federated Learning Architecture for Personalized Edge Intelligence

Zhiyuan Wu, Sheng Sun, Yuwei Wang, Min Liu, Wen Wang, Xuefeng Jiang, Bo Gao, Jinda Lu
IEEE Transactions on Mobile Computing (TMC)
arXiv

To tackle non-IID data issue in the context of Federated Learning with Knowledge Cache Architecture.

FedICT: Federated Multi-task Distillation for Multi-access Edge Computing

Zhiyuan Wu, Sheng Sun, Yuwei Wang, Min Liu,Xuefeng Jiang, Bo Gao
IEEE Transactions on Parallel and Distributed Systems (TPDS), 2023
arXiv

To deal with non-IID issue of Federated Learning with Knowledge Distillation.

FedBIAD: Communication-Efficient and Accuracy-Guaranteed Federated Learning with Bayesian Inference-Based Adaptive Dropout

Jingjing Xue, Min Liu, Sheng Sun, Yuwei Wang, Hui Jiang, Xuefeng Jiang
IEEE International Parallel and Distributed Processing Symposium (IPDPS), 2023
arXiv

To reduce communication cost in Federated Learning system via Adaptive Bayesian Dropout.

	FedTrip: A Resource-Efficient Federated Learning Method with Triplet Regularization Xujing Li, Min Liu, Sheng Sun, Yuwei Wang, Hui Jiang, Xuefeng Jiang IEEE International Parallel and Distributed Processing Symposium (IPDPS), 2023 arXiv To tackle non-IID data issue in Federated Learning with lower computation cost.
	Towards Federated Learning against Noisy Labels via Local Self-Regularization Xuefeng Jiang, Sheng Sun, Yuwei Wang, Min Liu ACM International Conference on Information & Knowledge Managements (CIKM), 2022 arXiv An early attempt to deal with noisy labels considering privacy concern of Federated Learning.

Image Segmentation Competition on Ultrasound image segmentation of hemangioma with only 200+ samples

Wenrui Liu, Xuefeng Jiang, Huan Chen, Junjie Luo, Yuan Wang
CCF BDCI 2021 Challenge on Ultrasound image segmentation of hemangioma
link

We (Flappy Pig) are the top 3 teams (175 teams in total) on this challenge. The evaluation metric is DICE value in Image Segmentation. Our DICE value is 0.902628848 (Top-1: 0.903214619). The base model is CaraNet updated by Adam optimizer.

Tech Reports & Preprints

VIPS-Odom: Visual-Inertial Odometry Tightly-coupled with Parking Slots for Autonomous Parking

Xuefeng Jiang, Fangyuan Wang, Rongzhang Zheng, Han Liu, Yixiong Huo, Jinzhang Peng, Lu Tian, Emad Barsoum
AMD's underground autonomous parking SLAM method.

Semantic Visual-inertial SLAM for Autonomous Parking

FNBench: Benchmarking Robust Federated Learning against Noisy Labels

Xuefeng Jiang, Jia Li, Nannan Wu, Zhiyuan Wu, Xujing Li, Sheng Sun, Gang Xu, Yuwei Wang, Qi Li, Min Liu
Submitted to IEEE Transactions on Dependable and Secure Computing
IEEE Tech Arxiv

To comprehensively benchmark sixteen existing methods on five datasets and inspect why noisy labels impair federated learning.

FedDSHAR: A Dual-StrategyFederated Learning Approach for Human Activity Recognition amid Noise Label User

Ziqian Lin, Xuefeng Jiang , Kun Zhang,Chongjun Fang, Liu yaya
Submitted to Future Generation Computer Systems and Received a major revision.

To deal with label noise in distributed HAR datasets.

FedLG: Using Two-Stage Sampling to Reduce The Impact of Class Imbalance and Label Noise on The Performance of Federated Learning

Ziqian Lin, Yaya Liu, Chongjun Fan, Liu Lei, Xuefeng Jiang
Submitted to Applied Intelligence

To deal with Non-IID and label noise in FL via sampling scheme.

DiffTimbre: Generate Timbre as You Describe through Latent Diffusion

Chenqi Lin, Xuefeng Jiang, Anlong Ming, Haiyang Zhang
Submitted to ECAI'24

To generate diverse speakers with different timbre with your prompt via latent diffusion.

Federated Skewed Label Learning with Logits Fusion

Yuwei Wang, Runhan Li, Hao Tan, Xuefeng Jiang, Sheng Sun, Min Liu, Bo Gao, Zhiyuan Wu
Arxiv

An explict yet effective method to exploit ensemble distillation to tackle Non-IID issue in Federated Learning.

FedSPL: Robust Federated Learning Against Noisy Labels via Self-Paced Learning

Zhiguo Da, Xuefeng Jiang, Yanming Chen
Submitted to Journal of Systems Architecture, received Major Revision

A follow-up work based on FedLSR.

Federated Class-Incremental Learning with New-Class Augmented Self-Distillation

Zhiyuan Wu, Tianliu He, Sheng Sun, Yuwei Wang, Min Liu, Bo Gao, Xuefeng Jiang
Submitted to ECAI'24

Incremental learning in FL with theoretical analysis.

Survey of knowledge distillation in federated edge learning

Zhiyuan Wu, Sheng Sun, Yuwei Wang, Min Liu, Xuefeng Jiang, Runhan Li, Bo Gao
Arxiv

Summary of federated distillation in edge computing environments.

🎓 Education

	PhD Candidate Period: 2021.09 - 2026.07 Institute of computing technology, Chinese Academy of Sciences (ICT, CAS) GPA: 3.9/4.0
	Degree: Bachelor Period: 2017.09 - 2021.07 Network Engineering, School of Computer Science, Beijing Posts and Telecommunications University (BUPT) GPA: 3.69/4.0 (Ranked 2 Out of 69)

Projects (to be updated)

Research Project with Huawei MindSpore Team about Model Ops & Federated Learning
AVP-SLAM: Auto Valet Parking.
Semantic VI-SLAM: We develop a real-time semantic SLAM framework by exploiting parking slots.
3D Perception Dataset: Collected 10k frames in beijing with real-time Lidar, Cameras, Object Annotations on Tesla Vehicle (Internal Use for AMD).
Awesome Federated Learning (1.1k+ stars on Github) as a main contributor
Image Manipulation Forensics with Segmentation model with related explaination
...

Work Experience

Algorithm R&D Co-op/intern @AMD focusing on AutoDrive&SLAM, 2022.10 - now
FedML open-source volunteer, 2022.3-2022.4
Backend intern @AI start-up, 2020.8 - 2020.10
QA intern @Yuanfudao, 2019.8 - 2019.10

Honors & Awards

National Scholarship, 2024
UCAS Merit Student and Outstanding student cadres, 2023
UCAS Second-Prize Academic Scholarship, 2023
ACM/CIKM NSF&SIGWeb Student Award, 2022
Merit Undergraduate Scholarship of Chinese Academy of Sciences, 2021
Beijing and BUPT Outstanding Graduates, 2021
BUPT Enterprise Scholarship (top 10 out of 600) offered by Fiberhome Communication Inc., 2020
Completion of National Undergraduate Innovation and Entrepreneurship Training Project and completed National Software Copyright Registration as project leader, 2019
National Mathematics Competition, Third Prize & National English Competition, Third Prize, 2018&2019
Mathematical Contest in Modeling, Honorable Prize, 2018&2019
BUPT Second-Prize Scholarship, 2018&2019

Super Idol: Iris [link]

Derived from Jon Barron's website.