Publication Type

Conference Proceeding Article

Version

acceptedVersion

Publication Date

4-2022

Abstract

Many deep learning applications, like keyword spotting [1], [2], require the incorporation of new concepts (classes) over time, referred to as Class Incremental Learning (CIL). The major challenge in CIL is catastrophic forgetting, i.e., preserving as much of the old knowledge as possible while learning new tasks. Various techniques, such as regularization, knowledge distillation, and the use of exemplars, have been proposed to resolve this issue. However, prior works primarily focus on the incremental learning step, while ignoring the optimization during the base model training. We hypothesise that a more transferable and generalizable feature representation from the base model would be beneficial to incremental learning.In this work, we adopt multitask learning during base model training to improve the feature generalizability. Specifically, instead of training a single model with all the base classes, we decompose the base classes into multiple subsets and regard each of them as a task. These tasks are trained concurrently and a shared feature extractor is obtained for incremental learning. We evaluate our approach on two datasets under various configurations. The results show that our approach enhances the average incremental learning accuracy by up to 5.5%, which enables more reliable and accurate keyword spotting over time. Moreover, the proposed approach can be combined with many existing techniques and provides additional performance gain.

Keywords

Class Incremental Learning, Continual Learning, Multitask Learning, Keyword Spotting

Discipline

Artificial Intelligence and Robotics | Computer Engineering

Research Areas

Intelligent Systems and Optimization

Publication

Proceedings of 2022 IEEE International Conference on Acoustics, Speech and Signal Processing, Singapore, May 22-27

First Page

1

Last Page

5

Identifier

10.1109/ICASSP43922.2022.9746862

Publisher

IEEE

City or Country

Singapore

Additional URL

http://doi.org/10.1109/ICASSP43922.2022.9746862

Share

COinS