ReLU (Rectified Linear Unit) is a nonlinear activation function defined as f(x) = max(0, x), commonly applied element-wise within the hidden layers of deep neural networks used in WiFi/CSI sensing tasks such as occupancy detection and crowd counting. It matters for the field because it introduces the nonlinearity necessary for neural networks to learn complex mappings between raw or processed CSI features and human presence or count estimates, while also mitigating the vanishing gradient problem to enable efficient training of deeper architectures. Key variants encountered in related sensing work include Leaky ReLU, which allows a small negative slope for inputs below zero to prevent dead neurons, and Parametric ReLU (PReLU), which learns the negative slope during training, both offering potential improvements in convergence and representational capacity over the standard formulation.
Source Papers
- A low-cost automatic people-counting system at bus stops using Wi-Fi probe requests and deep learning ↗ — A low-cost automatic people-counting system at bus stops usi
- BLE Can See: A Reinforcement Learning Approach for RF-based Indoor Occupancy Detection ↗ — BLE Can See: A Reinforcement Learning Approach for RF-based
- CRPF-QC: An Efficient CSI Recurrence Plot-Based Framework for Queue Counting ↗ — CRPF-QC: An Efficient CSI Recurrence Plot-Based Framework fo
- 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
- Investigation of Environment Dependence in Wi-Fi CSI-Based Crowd Counting Systems ↗ — Investigation of Environment Dependence in Wi-Fi CSI-Based C
- Towards Environment Independent Device Free Human Activity Recognition ↗ — Towards Environment Independent Device Free Human Activity R