Interpreting deep urban sound classification using Layer-wise Relevance Propagation
This work addresses the need for explainable AI in a sensitive application (assisting drivers with hearing loss), but it appears incremental as it applies existing interpretation methods to a specific domain.
The researchers tackled the problem of interpreting deep neural network predictions for urban sound classification, specifically for assisting drivers with hearing loss, by applying layer-wise relevance propagation to analyze model decisions using Mel and constant-Q spectrograms, identifying highly relevant frequency content as discriminative information.
After constructing a deep neural network for urban sound classification, this work focuses on the sensitive application of assisting drivers suffering from hearing loss. As such, clear etiology justifying and interpreting model predictions comprise a strong requirement. To this end, we used two different representations of audio signals, i.e. Mel and constant-Q spectrograms, while the decisions made by the deep neural network are explained via layer-wise relevance propagation. At the same time, frequency content assigned with high relevance in both feature sets, indicates extremely discriminative information characterizing the present classification task. Overall, we present an explainable AI framework for understanding deep urban sound classification.