Transfer Learning in Conversational Analysis through Reusing Preprocessing Data as Supervisors

TL;DR

This paper proposes reusing preprocessed data as auxiliary tasks in transfer learning for conversational analysis, improving performance and generalization over single-task methods.

Contribution

It introduces sixteen auxiliary tasks derived from preprocessing data, studies task capacity distribution, and explores supervision hierarchy to enhance conversational analysis models.

Findings

Improved accuracy over single-task models on IEMOCAP and SEMAINE datasets.

Identified effective auxiliary tasks that boost primary task performance.

Demonstrated potential for generalization across multiple primary tasks.

Abstract

Conversational analysis systems are trained using noisy human labels and often require heavy preprocessing during multi-modal feature extraction. Using noisy labels in single-task learning increases the risk of over-fitting. Auxiliary tasks could improve the performance of the primary task learning during the same training -- this approach sits in the intersection of transfer learning and multi-task learning (MTL). In this paper, we explore how the preprocessed data used for feature engineering can be re-used as auxiliary tasks, thereby promoting the productive use of data. Our main contributions are: (1) the identification of sixteen beneficially auxiliary tasks, (2) studying the method of distributing learning capacity between the primary and auxiliary tasks, and (3) studying the relative supervision hierarchy between the primary and auxiliary tasks. Extensive experiments on IEMOCAP and SEMAINE data validate the improvements over single-task approaches, and suggest that it may generalize across multiple primary tasks.