Research Overview
Our group investigates problems at the intersection of machine learning, sensing systems, and human-centered AI. Our research spans multiple domains with the shared goal of building robust, privacy-friendly, and adaptive intelligent systems.
Key Research Areas
-
🏠 Ambient Intelligence and Spatial Context Understanding in Smart Environments
We develop machine learning and sensing methods that enable smart environments to understand human activities, locations, and spatial context. Our research includes activity recognition from event-triggered sensors, layout-aware and trajectory-aware modeling, cross-modal sensor alignment, generative modeling for symbolic sensor streams, and indoor localization with wireless and mobile sensing.
Representative publications:
- LastAct: Trajectory-Guided Latest-Activity Localization for Real-Time Smart-Home Activity Recognition. arXiv'26
Z Liu*, R Fang*, J Lu, F Dou#.
- ADLGen: Synthesizing Symbolic, Event-Triggered Sensor Sequences for Human Activity Modeling. SenSys'26
W You*, H Jiang*, J Li*, Z Liu*, T Liu, J Lu, F Dou#.
- CARE: Contrastive Alignment for ADL Recognition from Event-Triggered Sensor Streams. PerCom'26
J Zhao, Z Liu, R Fang, J Lu, L Zhang, F Dou#.
- Cross-Modal Translation and Alignment of Sensor Events for Layout-Aware Activity Modeling. ISWC'25
W You, Z Liu, F Dou#.
- MARAuder's Map: Motion-Aware Real-time Activity Recognition with Layout-Based Trajectories. arXiv'25
Z Liu, W You, J Lu, F Dou#.
- On-Device Indoor Positioning: A Federated Reinforcement Learning Approach With Heterogeneous Devices. IoT-J'23
F Dou, J Lu, T Zhu, J Bi.
- A Bisection Reinforcement Learning Approach to 3-D Indoor Localization. IoT-J'20
F Dou, J Lu, T Xu, C-H Huang, J Bi.
- Top-Down Indoor Localization with Wi-Fi Fingerprints Using Deep Q-Network. MASS'18
F Dou, J Lu, Z Wang, X Xiao, J Bi, C-H Huang.
* Equal contribution. # Corresponding author.
🫀 Contactless Health Sensing and Physiological AI
We develop machine learning and sensing systems for contactless monitoring of human physiology in real-world environments. Our work spans problems such as blood pressure estimation, sleep apnea detection, respiration and snoring analysis, posture recognition, and physiological peak detection, with an emphasis on robustness, interpretability, and deployment in everyday and clinical settings.
Representative publications:
- DeepArrhythmia: Segment-Contextualized ECG Arrhythmia Classification via Selective Evidence Acquisition. arXiv'26
J Li, R Fang, Z Liu, W Song, J Lu, F Dou#.
- Cross-Subject Generalization for EEG Decoding: A Survey of Deep Learning Methods. PBME'26
T Li, Y Yan, F Dou, W Song, X Zhang.
- Peak-Detector: Explainable Peak Detection via Instruction-Tuned Large Language Models in Cardiac Physiological Signal. IMWUT'26
J Li, Y Zhang, Z Zeng, J Chen, Y Song, Y Xiao, N Dong, J Lu, Y Kwon, X Zhang, J Lu, W Song, F Dou#.
- Contactless Sleep Apnea Detection with Bodyseismography. IMWUT'26
Y Song*, J Chen*, Z Zeng, Y Zhang, Z F Pitafi, D K Das, B G Phillips, Y Kwon, X Zhang, F Dou, W Song.
- Attention Feature Fusion with Cluster Contrastive Learning for Snoring and Breath-Holding Detection Using Seismic Sensing. PerCom'26
Y Song*, J Chen*, Z Zeng, Y Zhang, Z F Pitafi, X Zhang, F Dou, W Song.
- Peak-R1: Instruction-Tuned Large Language Models for Robust J-Peak Detection in Cardiomechanical Signals. TS4H @ NeurIPS'25
J Li, Y Zhang, Z Zeng, J Chen, X Zhang, J Lu, W Song, F Dou#.
- Multi-granularity Supervised Contrastive Learning with Online Adaptation for Contactless In-bed Posture Classification. IMWUT'25
Y Song*, H Xiang*, Z Zeng, J Chen, Y Zhang, Z F Pitafi, H Yang, Q Lu, X Zhang, B G Phillips, F Dou, W Song.
- Proximal Federated Learning for Body Mass Index Monitoring using Commodity WiFi. PICASSO @ MobiCom'24
J Li, K Davuluri, K Mottakin, Z Song, F Dou, J Lu.
- Self-Supervised Representation Learning and Temporal-Spectral Feature Fusion for Bed Occupancy Detection. IMWUT'24
Y Song, Z F Pitafi, F Dou, J Sun, X Zhang, B G Phillips, W Song.
* Equal contribution. # Corresponding author.
🧠 Foundation Models, Representation Learning, and Scientific AI
We investigate modern machine learning methods, including foundation models, contrastive learning, cross-modal understanding, model merging, efficient LLM fine-tuning, representation learning, computer vision, biomedical AI, and scientific machine learning.
Representative publications:
- LARV: Data-Free Layer-wise Adaptive Rescaling Veneer for Model Merging. arXiv'26
X Wang, K Deng, F Dou, J Bi, J Lu.
- SIACL: Structure-Invariant Autoregressive Contrastive Learning for Generative Chemical Language Models via Isomorphic Trajectory Alignment. arXiv'26
X Wang, F Dou, J Bi, M Song.
- Achieving Fine-grained Cross-modal Understanding through Brain-inspired Hierarchical Representation Learning. arXiv'26
W You, H Jiang, Y Pan, J Chen, T Liu, F Dou#.
- HELENE: Hessian Layer-wise Clipping and Gradient Annealing for Accelerating Fine-tuning LLM with Zeroth-order Optimization. EMNLP'25
H Zhao*, J Li*, Y Pan, S Liang, X Yang, F Dou, T Liu, J Lu.
- LOCAL: Latent Orthonormal Contrastive Learning for Paired Image Classification. ICCV Findings'25 Oral
F Dou, J Lu, T Zhu, J Bi.
- Polyhedron Attention Module: Learning Adaptive-order Interactions. NeurIPS'23
T Zhu, F Dou, X Wang, J Lu, J Bi.
- Customized Positional Encoding to Combine Static and Time-varying Data in Robust Representation Learning for Crop Yield Prediction. IJCAI'23
Q Liu, F Dou, M Yang, E Amdework, G Wang, J Bi.
⚙️ Distributed, Federated, and Adaptive Machine Learning
We design learning and optimization methods for decentralized, heterogeneous, infrastructure-less, and resource-constrained environments. Our work addresses federated learning, personalized optimization, federated reinforcement learning, and adaptive decision-making for networked and sensing systems.
Representative publications:
- Deep Q-Learning-Based Mobile Charger Path Planning in Wireless Powered Communication Networks. TECS'25
M Mondal, F Dou, J Bi, S Han.
- Proximal Federated Learning for Body Mass Index Monitoring using Commodity WiFi. PICASSO @ MobiCom'24
J Li, K Davuluri, K Mottakin, Z Song, F Dou, J Lu.
- Mobilizing Personalized Federated Learning in Infrastructure-Less and Heterogeneous Environments via Random Walk Stochastic ADMM. NeurIPS'23
Z Parsons, F Dou, H Du, Z Song, J Lu.
- On-Device Indoor Positioning: A Federated Reinforcement Learning Approach With Heterogeneous Devices. IoT-J'23
F Dou, J Lu, T Zhu, J Bi.
🌊 Earlier Work in Underwater Sensing and Networking Systems
Our earlier work focused on underwater wireless sensor networks, including secure localization, underwater acoustic communication, MAC protocols, network simulation, and sensing/communication in challenging environments.
Representative publications:
- Secure Localization for Underwater Wireless Sensor Networks via AUV Cooperative Beamforming with Reinforcement Learning. TMC'24
R Fan, A Boukerche, P Pan, Z Jin, Y Su, F Dou.
- A Secure Localization Scheme for UWSNs Based on AUV Formation Cooperative Beamforming. ICC'24
R Fan, P Pan, Z Jin, Y Su, F Dou.
- A Secure Communication Scheme Based on Spatio-Temporal Dynamics of Underwater Acoustic Channel. ICC'24
Y Su, P Pan, R Fan, S Yang, F Dou.
- Aqua-Sim Next Generation: A NS-3 Based Simulator for Underwater Sensor Networks. WUWNet'15
R Martin, Y Zhu, L Pu, F Dou, Z Peng, J-H Cui, S Rajasekaran.
- On-Demand Pipelined MAC for Multi-Hop Underwater Wireless Sensor Networks. WUWNet'15
F Dou, Z Peng.
- The Multi-Channel MAC Protocol for High-Performance Underwater Sensor Networks. (in Chinese) JHEU'15
Y Su, Z Jin, F Dou.
- Motion Prediction Based MAC for Underwater Wireless Sensor Networks. (in Chinese) JEIT'13
Z Jin, Y Su, Z Liu, F Dou#.
- WSF-MAC: A Weight-Based Spatially Fair MAC Protocol for Underwater Sensor Networks. CECNet'12
F Dou, Z Jin, Y Su, J Liu.
- Nodes Distribution Method Aiming at Improving the Fairness for Underwater Acoustic 3D Sensor Networks. Patent
Z Jin, Z Liu, F Dou. CN103095382B.
- Spatially Fair Media Access Control Method for Underwater Sensor Networks. Patent
Z Jin, F Dou, Y Su. CN102612091B.