Investigation of Feature Selection and Pooling Methods for Environmental Sound Classification

TL;DR

This study evaluates feature selection and pooling strategies, especially Sparse Salient Region Pooling, for environmental sound classification using lightweight CNNs, demonstrating significant accuracy improvements over traditional methods.

Contribution

The paper introduces and assesses SSRP variants for ESC, showing their effectiveness over PCA and baseline CNNs in resource-limited environments.

Findings

SSRP-T achieves up to 80.69% accuracy on ESC-50

SSRP methods outperform PCA and baseline CNNs

Sparse pooling enhances efficiency and robustness

Abstract

This paper explores the impact of dimensionality reduction and pooling methods for Environmental Sound Classification (ESC) using lightweight CNNs. We evaluate Sparse Salient Region Pooling (SSRP) and its variants, SSRP-Basic (SSRP-B) and SSRP-Top-K (SSRP-T), under various hyperparameter settings and compare them with Principal Component Analysis (PCA). Experiments on the ESC-50 dataset demonstrate that SSRP-T achieves up to 80.69 % accuracy, significantly outperforming both the baseline CNN (66.75 %) and the PCA-reduced model (37.60 %). Our findings confirm that a well-tuned sparse pooling strategy provides a robust, efficient, and high-performing solution for ESC tasks, particularly in resource-constrained scenarios where balancing accuracy and computational cost is crucial.