Learning an Evolved Mixture Model for Task-Free Continual Learning

Fei Ye; Adrian Gheorghe Bors

doi:10.1109/ICIP46576.2022.9898047

Learning an Evolved Mixture Model for Task-Free Continual Learning

Fei Ye, Adrian Gheorghe Bors

Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Recently, continual learning (CL) has gained significant interest because it enables deep learning models to acquire new knowledge without forgetting previously learnt information. However, most existing works require knowing the task identities and boundaries, which is not realistic in a real context. In this paper, we address a more challenging and realistic setting in CL, namely the Task-Free Continual Learning (TFCL) in which a model is trained on non-stationary data streams with no explicit task information. To address TFCL, we introduce an evolved mixture model whose network architecture is dynamically expanded to adapt to the data distribution shift. We implement this expansion mechanism by evaluating the probability distance between the knowledge stored in each mixture model component and the current memory buffer using the Hilbert Schmidt Independence Criterion (HSIC). We further introduce two simple dropout mechanisms to selectively remove stored examples in order to avoid memory overload while preserving memory diversity. Empirical results demonstrate that the proposed approach achieves excellent performance.

Original language	English
Title of host publication	IEEE International Conference on Image Processing (ICIP)
Place of Publication	Bordeaux, France
Publisher	IEEE
Pages	1936-1940
ISBN (Print)	9781665496209
DOIs	https://doi.org/10.1109/ICIP46576.2022.9898047
Publication status	Published - 18 Oct 2022

Bibliographical note

© 2022 IEEE. This is an author-produced version of the published paper. Uploaded in accordance with the publisher’s self-archiving policy. Further copying may not be permitted; contact the publisher for details

Access to Document

10.1109/ICIP46576.2022.9898047

Learning an Evolved Mixture Model for Task-Free Continual Learning
© 2022 IEEE. This is an author-produced version of the published paper. Uploaded in accordance with the publisher’s self-archiving policy. Further copying may not be permitted; contact the publisher for details
Accepted author manuscript, 1.37 MB

Cite this

@inproceedings{846b6d0b8a304b48ba85aee33b88bfe9,

title = "Learning an Evolved Mixture Model for Task-Free Continual Learning",

abstract = "Recently, continual learning (CL) has gained significant interest because it enables deep learning models to acquire new knowledge without forgetting previously learnt information. However, most existing works require knowing the task identities and boundaries, which is not realistic in a real context. In this paper, we address a more challenging and realistic setting in CL, namely the Task-Free Continual Learning (TFCL) in which a model is trained on non-stationary data streams with no explicit task information. To address TFCL, we introduce an evolved mixture model whose network architecture is dynamically expanded to adapt to the data distribution shift. We implement this expansion mechanism by evaluating the probability distance between the knowledge stored in each mixture model component and the current memory buffer using the Hilbert Schmidt Independence Criterion (HSIC). We further introduce two simple dropout mechanisms to selectively remove stored examples in order to avoid memory overload while preserving memory diversity. Empirical results demonstrate that the proposed approach achieves excellent performance.",

author = "Fei Ye and Bors, {Adrian Gheorghe}",

note = "{\textcopyright} 2022 IEEE. This is an author-produced version of the published paper. Uploaded in accordance with the publisher{\textquoteright}s self-archiving policy. Further copying may not be permitted; contact the publisher for details",

year = "2022",

month = oct,

day = "18",

doi = "10.1109/ICIP46576.2022.9898047",

language = "English",

isbn = "9781665496209",

pages = "1936--1940",

booktitle = "IEEE International Conference on Image Processing (ICIP)",

publisher = "IEEE",

address = "United States",

}

TY - GEN

T1 - Learning an Evolved Mixture Model for Task-Free Continual Learning

AU - Ye, Fei

AU - Bors, Adrian Gheorghe

N1 - © 2022 IEEE. This is an author-produced version of the published paper. Uploaded in accordance with the publisher’s self-archiving policy. Further copying may not be permitted; contact the publisher for details

PY - 2022/10/18

Y1 - 2022/10/18

N2 - Recently, continual learning (CL) has gained significant interest because it enables deep learning models to acquire new knowledge without forgetting previously learnt information. However, most existing works require knowing the task identities and boundaries, which is not realistic in a real context. In this paper, we address a more challenging and realistic setting in CL, namely the Task-Free Continual Learning (TFCL) in which a model is trained on non-stationary data streams with no explicit task information. To address TFCL, we introduce an evolved mixture model whose network architecture is dynamically expanded to adapt to the data distribution shift. We implement this expansion mechanism by evaluating the probability distance between the knowledge stored in each mixture model component and the current memory buffer using the Hilbert Schmidt Independence Criterion (HSIC). We further introduce two simple dropout mechanisms to selectively remove stored examples in order to avoid memory overload while preserving memory diversity. Empirical results demonstrate that the proposed approach achieves excellent performance.

AB - Recently, continual learning (CL) has gained significant interest because it enables deep learning models to acquire new knowledge without forgetting previously learnt information. However, most existing works require knowing the task identities and boundaries, which is not realistic in a real context. In this paper, we address a more challenging and realistic setting in CL, namely the Task-Free Continual Learning (TFCL) in which a model is trained on non-stationary data streams with no explicit task information. To address TFCL, we introduce an evolved mixture model whose network architecture is dynamically expanded to adapt to the data distribution shift. We implement this expansion mechanism by evaluating the probability distance between the knowledge stored in each mixture model component and the current memory buffer using the Hilbert Schmidt Independence Criterion (HSIC). We further introduce two simple dropout mechanisms to selectively remove stored examples in order to avoid memory overload while preserving memory diversity. Empirical results demonstrate that the proposed approach achieves excellent performance.

U2 - 10.1109/ICIP46576.2022.9898047

DO - 10.1109/ICIP46576.2022.9898047

M3 - Conference contribution

SN - 9781665496209

SP - 1936

EP - 1940

BT - IEEE International Conference on Image Processing (ICIP)

PB - IEEE

CY - Bordeaux, France

ER -