DNN Transfer Learning based Non-linear Feature Extraction for Acoustic Event Classification

TL;DR

This paper introduces a DNN-based non-linear feature extraction method for acoustic event classification, improving performance and robustness in indoor surveillance scenarios with limited data.

Contribution

It presents a novel DNN transfer learning approach for non-linear feature extraction, addressing limitations of traditional linear filters in acoustic event classification.

Findings

Enhanced classification accuracy demonstrated

Robustness to noise confirmed

Effective with limited target data

Abstract

Recent acoustic event classification research has focused on training suitable filters to represent acoustic events. However, due to limited availability of target event databases and linearity of conventional filters, there is still room for improving performance. By exploiting the non-linear modeling of deep neural networks (DNNs) and their ability to learn beyond pre-trained environments, this letter proposes a DNN-based feature extraction scheme for the classification of acoustic events. The effectiveness and robustness to noise of the proposed method are demonstrated using a database of indoor surveillance environments.