Continual Learning for Affective Computing

TL;DR

This paper explores using continual learning to develop personalized affect perception models that adapt to individual differences in emotional expression, addressing limitations of current benchmark-focused approaches.

Contribution

It introduces continual learning as a novel paradigm for personalized affective computing, enhancing model adaptation to individual differences.

Findings

Continual learning improves personalization in affective models.

Current benchmarks do not capture individual variability.

Proposed approach enhances affect recognition accuracy.

Abstract

Real-world application requires affect perception models to be sensitive to individual differences in expression. As each user is different and expresses differently, these models need to personalise towards each individual to adequately capture their expressions and thus, model their affective state. Despite high performance on benchmarks, current approaches fall short in such adaptation. In this work, we propose the use of Continual Learning (CL) for affective computing as a paradigm for developing personalised affect perception.