Continual Learning in Practice

Tom Diethe; Tom Borchert; Eno Thereska; Borja Balle; Neil Lawrence

edit

Back to publications

Continual Learning in Practice

Tom Diethe, Tom Borchert, Eno Thereska, Borja Balle, Neil Lawrence

, 2019.

Abstract

This paper describes a reference architecture for self-maintaining systems that can learn continually, as data arrives. In environments where data evolves, we need architectures that manage Machine Learning (ML) models in production, adapt to shifting data distributions, cope with outliers, retrain when necessary, and adapt to new tasks. This represents continual AutoML or Automatically Adaptive Machine Learning. We describe the challenges and proposes a reference architecture.

Links

Cite this Paper

BibTeX


@Misc{continual-learning-in-practice,
  title = 	 {Continual Learning in Practice},
  author = 	 {Diethe, Tom and Borchert, Tom and Thereska, Eno and Balle, Borja and Lawrence, Neil},
  year = 	 {2019},
  url = 	 {/publications/continual-learning-in-practice.html},
  abstract = 	 {This paper describes a reference architecture for self-maintaining systems that can learn continually, as data arrives. In environments where data evolves, we need architectures that manage Machine Learning (ML) models in production, adapt to shifting data distributions, cope with outliers, retrain when necessary, and adapt to new tasks. This represents continual AutoML or Automatically Adaptive Machine Learning. We describe the challenges and proposes a reference architecture.
}
}

Endnote

%0 Generic
%T Continual Learning in Practice
%A Tom Diethe
%A Tom Borchert
%A Eno Thereska
%A Borja Balle
%A Neil Lawrence
%D 2019	
%F continual-learning-in-practice
%U /publications/continual-learning-in-practice.html
%X This paper describes a reference architecture for self-maintaining systems that can learn continually, as data arrives. In environments where data evolves, we need architectures that manage Machine Learning (ML) models in production, adapt to shifting data distributions, cope with outliers, retrain when necessary, and adapt to new tasks. This represents continual AutoML or Automatically Adaptive Machine Learning. We describe the challenges and proposes a reference architecture.

RIS


TY  - GEN
TI  - Continual Learning in Practice
AU  - Tom Diethe
AU  - Tom Borchert
AU  - Eno Thereska
AU  - Borja Balle
AU  - Neil Lawrence
DA  - 2019/03/12	
ID  - continual-learning-in-practice
UR  - /publications/continual-learning-in-practice.html
AB  - This paper describes a reference architecture for self-maintaining systems that can learn continually, as data arrives. In environments where data evolves, we need architectures that manage Machine Learning (ML) models in production, adapt to shifting data distributions, cope with outliers, retrain when necessary, and adapt to new tasks. This represents continual AutoML or Automatically Adaptive Machine Learning. We describe the challenges and proposes a reference architecture.

ER  -

APA


Diethe, T., Borchert, T., Thereska, E., Balle, B. & Lawrence, N.. (2019). Continual Learning in Practice.  Available from /publications/continual-learning-in-practice.html.