Research

Current Research Topics

• AirComp (Over-the-air computing)
• (preprint) Communication-Efficient Federated Learning via Synchronous Averaging
• Non-Terrestrial Network (NTN)
• (TWC 2024) Non-iterative Optimization of Trajectory and Radio Resource for Aerial Network
• Communication-efficient optimization
• (ICML 2024) Rethinking DP-SGD in Discrete Domain: Exploring Logistic Distribution in the Realm of signSGD
• (TVT 2023) Distributed Resource Allocation and User Association for Max-Min Fairness in HetNets
• Trustworthy AI
• (preprint) TA-VLM: Generating Targeted Adversarial Examples for Vision-Language Models
• (preprint) Data Reconstruction Attack Against Adversarial Visual Information Hiding
• (preprint) Deeper Understanding of Black-box Predictions via Generalized Influence Functions
• (preprint) Patch-MI: Enhancing Model Inversion Attacks via Patch-Based Reconstruction

Previous Research Topics

• Radio resource management
• (TMC 2024) Age-of-Information-Aware Distributed Task Offloading and Resource Allocation in Mobile Edge
• (TWC 2022) α-Fairness Maximizing User Association in Energy-Constrained Small Cell Networks
• (TCOM 2022) Deep Learning-Aided User Association and Power Control with Renewable Energy Sources
• (TVT 2022) Recurrent Neural Network-Based User Association and Power Control in Dynamic HetNets
• (TVT 2020) Deep Reinforcement Learning-based Resource Allocation and Power Control in Small Cells with Limited Information Exchange
• (TVT 2019) Resource Allocation and Power Control in Cooperative Small Cell Networks with Backhaul Constraint
• AI/ML applications in real-world problems
• (NeurIPS 2023) M²SODAI: Multi-Modal Maritime Object Detection Dataset With RGB and Hyperspectral Image Sensors
• (IROS 2020) Deep Learning-Based Autonomous Scanning Electron Microscope

Useful papers

1. Measuring Data Influnece

Title	Team/Authors	Venue/Year	Note
Counterfactual Memorization in Neural Language Models	Chiyuan Zhang et. al.	NeurIPS 2023	A low-complexity method for measuring impact of a data (or a subset)

2. Optimization Papers

Title	Team/Authors	Venue/Year	Note
signSGD with Majority Vote is Communication Efficient And Fault Tolerant	Jeremy Bernstein	ICLR 2019	A majority voting signSGD is proposed, which is more Fault Tolerant compared to baseline signSGD

3. Differential Privacy

Title	Team/Authors	Venue/Year	Note
Additive Logistic Mechanism for Privacy-Preserving Self-Supervised Learning	Yunhao Yang, et. al.	arxiv 2022	This paper proposes a new additive noise mechanism: Logistic mechanism.

4. Adversarial Example

Title	Team/Authors	Venue/Year	Note
Hiding Visual Information via Obfuscating Adversarial Perturbations	Zhigang Su, et. al.	ICCV 2023	hiding visual information via type-I adversarial attack

5. Federated Learning

Title	Team/Authors	Venue/Year	Note
Federated Unlearning via Active Forgetting	Yuyuan Li, et. al.	arxiv 2023	This paper proposes a federated unlearning framework.
One-shot Empirical Privacy Estimation for Federated Learning	Galen Andrew, et. el.	arxiv 2023	This paper introduces a canary-based DP auditing method for FL environment (slides)
SplitFed: When Federated Learning Meets Split Learning	Chandra Thapa, et. al.	AAAI 2022	This paper proposes a combination of FL and SL
One the Convergence of FedAvg on Non-IID Data	Xiang Li, et. al.	ICLR 2020	This paper proves the convergence of FedAvg on Non-IID data (full/partial participation)
Communication-Efficient Learning of Deep Networks from Decentralized Data	Google	AISTATS 2017	This paper proposes FedAvg
Adaptive Federated Learning in Resource Constrained Edge Computing Systems	Shiqiang Wang, et .al.	IEEE JSAC 2019	Investigates how the number of update per commun. round affect the convergence.