An Attention-based Approach to Hierarchical Multi-label Music Instrument Classification

TL;DR

This paper introduces a hierarchical multi-label music instrument classification task and proposes two novel joint training methods using rule-based pooling and attention mechanisms, demonstrating improved performance and interpretability.

Contribution

It presents a new hierarchical multi-label classification framework for music instruments and introduces two innovative joint training techniques utilizing rule-based and attention-based models.

Findings

Proposed methods outperform non-joint training approaches.

Attention-based method offers interpretability through attention maps.

Joint training improves classification accuracy.

Abstract

Although music is typically multi-label, many works have studied hierarchical music tagging with simplified settings such as single-label data. Moreover, there lacks a framework to describe various joint training methods under the multi-label setting. In order to discuss the above topics, we introduce hierarchical multi-label music instrument classification task. The task provides a realistic setting where multi-instrument real music data is assumed. Various hierarchical methods that jointly train a DNN are summarized and explored in the context of the fusion of deep learning and conventional techniques. For the effective joint training in the multi-label setting, we propose two methods to model the connection between fine- and coarse-level tags, where one uses rule-based grouped max-pooling, the other one uses the attention mechanism obtained in a data-driven manner. Our evaluation reveals that the proposed methods have advantages over the method without joint training. In addition, the decision procedure within the proposed methods can be interpreted by visualizing attention maps or referring to fixed rules.