Quantifying Uncertainty in Real Time with Split BiRNN for Radar Human Activity Recognition

Radar systems can be used to perform human activity recognition in a privacy preserving manner. Deep Neural Networks are able to effectively process the complex radar data and make predictions. Often these networks are large and do not scale well when processing a large amount of radar streams at once, for example when monitoring multiple rooms in a hospital. This work proposes Bayesian Split Bidirectional Recurrent Neural Network for Human Activity Recognition. Using this technique the processing of data is split in two parts, one part on-premise (low-power, low-cost device), and the other off-premise (high power device). The proposed approach leverages the power of the off-premise device to quantify its uncertainty, and to gain more information on its epistemic and its aleatoric parts. Results indicate the proposed approach is able to correctly identify parts of a prediction that either need more training data for better predictions (epistemic uncertainty), or are inherently hard to classify by the model (aleatoric uncertainty).

I. Introduction

Real-time activity recognition in a hospital room or nursing home is important, because it can help to detect troublesome events, such as the fall of a patient, as soon as possible. This is most meaningful for geriatric patients [1], [2] that are more likely to suffer lasting injuries from a fall, especially if treatment is delayed. Using radar, a privacy-preserving method to detect falls can be established. A popular technique to perform fall detection, or human activity recognition is Deep Learning (DL). Data-driven DL models require high training times and powerful machines at prediction time. When this is deployed at scale to handle many radars at once, the cost of using DL increases significantly.