Songze Li

songzeli8824 [at] gmail [dot] com

songzeli [at] ust [dot] hk

Google Scholar
Github

Songze Li

Assistant Professor

Internet of Things Thrust
Artificial Intelligence Thrust
The Hong Kong University of Science and Technology (Guangzhou)

Affiliate Assistant Professor

Department of Computer Science and Engineering
The Hong Kong University of Science and Technology

Positions are open for strongly motivated postgraduate students (MPhil and PhD), post-docs, research assistants, and visiting interns, on research topics including federated learning algorithms, security, and privacy, secure multiparty computation (e.g., threshold cryptosystems and secure ML protocols), and blockchain security and scalability (e.g., sharding and rollups).
Interested applicants please email me with your CV, transcript, and any related publications at songzeli [at] ust [dot] hk. Past experience on publishing (as first author) on top machine learning, computer/communciation security, cryptography, and information theory venues is a big plus!

Research Interests

AI security and privacy

Secure multi-party computation

Blockchain security and scalability

News

[April 29, 2023] Our paper "Communication-Efficient Coded Computing for Distributed Multi-Task Learning" is accepted to IEEE Transactions on Communications!

[Apr. 25, 2023] Two papers on federated learning security and privacy are accepted to ICML 2023!
- FedVS: Straggler-Resilient and Privacy-Preserving Vertical Federated Learning for Split Models. Congratulations, Duanyi and Jin!
- Chameleon: Adapting to Peer Images for Planting Durable Backdoors in Federated Learning. Congratulations, Yanbo!

[Apr. 15, 2023] Two papers on secure federated learning are acceped to ISIT 2023!
- Secure Embedding Aggregation for Federated Representation Learning. Congratulations, Jiaxiang!
- Secure Gradient Aggregation for Wireless Multi-Server Federated Learning

[Feb. 25, 2023] Our paper "Coded Distributed Computing for Hierarchical Multi-Task Learning" is accepted to IEEE Information Theory Workshop (ITW) 2023!

[Feb. 16, 2023] Our paper "Information-Theoretically Private Matrix Multiplication From MDS-Coded Storage" is accepted for publication on IEEE Transactions on Information Forensics & Security！

[Nov. 30, 2022] Our paper "SwiftAgg+: Achieving Asymptotically Optimal Communication Loads in Secure Aggregation for Federated Learning" is accepted for publication on IEEE Journal on Selected Areas in Communications (JSAC) 2022 Special Issue on Communication-Efficient Distributed Learning over Networks！

[Sept. 15, 2022] Our paper "DReS-FL: Dropout-Resilient Secure Federated Learning for Non-IID Clients via Secret Data Sharing" is accepted to NeurIPS 2022!

[Aug. 12, 2022] I am guest editing a special issue on Information Theory for Distributed Systems with Entropy. We look forward to receiving your great research works on coding for distributed storage systems, coded caching networks, coded distributed computation, distributed and federated learning systems, distributed private information and function retrieval, distributed ledgers/blockchains, and secure multi-party computation and verifiable computation.

[June 3, 2022] Our paper "A Systematic Approach towards Efficient Private Matrix Multiplication" is accepted to the Special Issue on Distributed Coding and Computation in the IEEE Journal on Selected Areas in Information Theory (JSAIT)!

[Apr. 23, 2022] Three papers on coded computing and federated learning are acceped to ISIT 2022!
- Generalized Lagrange Coded Computing: A Flexible Computation-Communication Tradeoff
- Stochastic Coded Federated Learning With Convergence and Privacy Guarantees
- SwiftAgg: Communication-Efficient and Dropout-Resistant Secure Aggregation for Federated Learning With Worst-Case Security Guarantees

[Jan. 14, 2022] Our paper "LightSecAgg: a Lightweight and Versatile Design for Secure Aggregation in Federated Learning" is accepted to Fifth Conference on Machine Learning and Systems (MLSys), 2022!

[Sept. 24, 2021] Our paper "LightSecAgg: Rethinking Secure Aggregation in Federated Learning" is accepted to 2021 IEEE Information Theory Workshop (ITW 2021).

[Aug. 2, 2021] Early admission to 2022 PhD and MPhil in IoT at HKUST(GZ) is now open! To apply, please send your application materials to iott@ust.hk.

[July 12, 2021] Our paper "OmniLytics: A Blockchain-based Secure Data Market for Decentralized Machine Learning" is accepted to ICML International Workshop on Federated Learning for User Privacy and Data Confidentiality (FL-ICML'21).

[May 21, 2021] I delivered an invited talk "DeepPrivate: Scalable Distributed DNN Training with Data and Model Privacy" at Sixth London Symposium on Information Theory (LSIT 2021).
[Paper]

[May 20, 2021] I delivered an invited talk "Nakamoto Meets Shannon: Scaling Blockchains Using Codes" at Iran Workshop on Communication and Information Theory (IWCIT 2021).

[May 13, 2021] We currently have multiple post-doc positions open. Applicants with a PhD degree who have background on one or more of the following research topics are encouraged to apply:
- Distributed Machine Learning algorithms and frameworks;
- Distributed Machine Learning systems (experience with ML frameworks like TensorFlow and PyTorch);
- Reinforcement Learning with applications in mission critical IoT and Networked Control systems;
- Modeling of Dynamic Systems;
- Stochastic Optimization and Analysis with focus on wireless communications and IoT systems.
Salary is highly competitive of international standard. Interested appliants please send application materials to me at songzeli [at] ust [dot] hk or Professor Danny Tsang at eetsang [at] ust [dot] hk.

[Apr. 14, 2021] One post-doc position is open in my group. Topics include distributed machine learning and blockchain systems. Salary is highly competitive of international standard. Successful applicants are entitled to additional subsidies for living and research. Interested appliants please send application materials directly to me at songzeli [at] ust [dot] hk.

[Feb. 9, 2021] Our paper "FedML: A Research Library and Benchmark for Federated Machine Learning" (a shorter version of FedML whitepaper) won the Baidu Best Paper Award at NeurIPS-20 Workshop on Scalability, Privacy, and Security in Federated Learning.

[Jan. 16, 2021] Our paper "Compressed Coded Distributed Computing" is accepted for publication in the IEEE Transactions on Communications.

[Nov. 24, 2020] Excited to introduce TaiJi, the first State Machine Replication (or blockchain) consensus protocol that simultaneously achieves Bitcoin dynamic availability and unpredictability, and fast confirmation latency of BFT protocols, with provable security against fully-adaptive adversary with up to 50% hashing power.
[arXiv] [IACR e-print] [zk Capital This Week in Blockchain Research Issue #84]

[Nov. 24, 2020] A shorter version of FedML whitepaper is accepted to NeurIPS 2020 SpicyFL Workshop and nominated for best paper award!

[Aug. 2020] Together with amazing collaborators from academia and industry, we release a software library FedML for federated learning, which allows fast prototying and experimenting research ideas across various computing environments. Please give it a try and your feedbacks are greatly appreciated.

[Aug. 2020] Together with my PhD advisor Salman Avestimehr, we publish a monograph on coded computing.

[July 2020] Our paper "PolyShard: Coded Sharding Achieves Linearly Scaling Efficiency and Security Simultaneously" is published on IEEE Transactions on Information Forensics and Security. Congratulations Fisher, Chien-Sheng, Salman, Sreeram, and Pramod!