Transfer learning is a machine learning technique in which knowledge gained from training a model on a source domain (e.g., a specific environment, person, or hardware configuration) is leveraged to improve performance on a different but related target domain, typically with limited labeled data. In Wi-Fi and CSI-based sensing, transfer learning is critical for addressing the generalizability problem, where models trained in one setting often fail when deployed in new locations, with unseen users, or using different antenna configurations due to the highly environment-dependent nature of channel state information. Key variants employed in this field include domain adaptation, which aligns feature distributions between source and target domains, fine-tuning, which partially retrains a pre-trained model on small amounts of target-domain data, and domain-adversarial training, which explicitly learns domain-invariant representations to reduce sensitivity to deployment-specific variations.
Source Papers
- CrossSense: Towards Cross-Site and Large-Scale WiFi Sensing ↗ — CrossSense: Towards Cross-Site and Large-Scale WiFi Sensing
- DASECount: Domain-Agnostic Sample-Efficient Wireless Indoor Crowd Counting via Few-Shot Learning ↗ — DASECount: Domain-Agnostic Sample-Efficient Wireless Indoor
- Deep Learning-Enhanced Human Sensing with Channel State Information: A Survey ↗ — Deep Learning-Enhanced Human Sensing with Channel State Info
- Occupancy Prediction in IoT-Enabled Smart Buildings: Technologies, Methods, and Future Directions ↗ — Occupancy Prediction in IoT-Enabled Smart Buildings: Technol
- Recent trends in crowd analysis: A review ↗ — Recent trends in crowd analysis: A review
- SenseFi: A library and benchmark on deep-learning-empowered WiFi human sensing ↗ — SenseFi: A library and benchmark on deep-learning-empowered
- Time matters: Empirical insights into the limits and challenges of temporal generalization in CSI-based Wi-Fi sensing ↗ — Time matters: Empirical insights into the limits and challen
- Towards Environment Independent Device Free Human Activity Recognition ↗ — Towards Environment Independent Device Free Human Activity R
- WiGNN: WiFi-Based Cross-Domain Gesture Recognition Inspired by Dynamic Topology Structure ↗ — WiGNN: WiFi-Based Cross-Domain Gesture Recognition Inspired