t-SNE (t-distributed Stochastic Neighbor Embedding) is a nonlinear dimensionality reduction technique that projects high-dimensional data into a low-dimensional space (typically 2D or 3D) by preserving local neighborhood structure, making it well-suited for visualizing complex feature distributions in CSI-based sensing systems. In Wi-Fi sensing research, it is primarily used as a diagnostic and interpretability tool to assess the quality of learned feature representations, for instance by revealing how well domain-adaptive embeddings such as those produced by PULSE or CNN-extracted CSI features cluster according to activity or occupancy class. While t-SNE does not produce variants in the strict algorithmic sense, it is commonly applied with different perplexity and iteration settings to balance local versus global structure preservation, and is often contrasted with related methods such as UMAP or PCA when evaluating the separability of learned sensing features across domains or environments.
Source Papers
- Channel State Information (CSI) Amplitude Coloring Scheme for Enhancing Accuracy of an Indoor Occupancy Detection System Using Wi-Fi Sensing ↗ — Channel State Information (CSI) Amplitude Coloring Scheme fo
- Context-Aware Predictive Coding: A Representation Learning Framework for WiFi Sensing ↗ — Context-Aware Predictive Coding: A Representation Learning F
- Efficient machine learning for Wi-Fi CSI-based human activity recognition using fast Monte Carlo based feature extraction ↗ — Efficient machine learning for Wi-Fi CSI-based human activit
- PULSE: Physics-Aware Temporal Embedding Learning for Domain Adaptive Wireless Sensing ↗ — PULSE: Physics-Aware Temporal Embedding Learning for Domain
- Towards Environment Independent Device Free Human Activity Recognition ↗ — Towards Environment Independent Device Free Human Activity R