Towards a Unified Lifelong Learning Framework

Tanner A. Bohn, Xinyu Yun, Charles X. Ling
NeurIPS 2020 Workshop on Pre-registration in Machine Learning, PMLR 148:221-235, 2021.

Abstract

Humans can learn a variety of concepts and skills incrementally over the course of their lives while exhibiting many desirable properties, such as continual learning without forgetting, forward transfer of knowledge, and learning a new concept with few examples. However, most previous approaches to efficient lifelong learning demonstrate only subsets of these properties, often by different complex mechanisms. In this preregistration submission, we propose to study the effectiveness of a unified lifelong learning framework designed to achieve many of these properties through one central mechanism. We describe this consolidation-based approach and propose experimental protocols to benchmark it on several skills, using grid searches over hyperparameters to better understand the framework.

Cite this Paper


BibTeX
@InProceedings{pmlr-v148-bohn21a, title = {Towards a Unified Lifelong Learning Framework}, author = {Bohn, Tanner A. and Yun, Xinyu and Ling, Charles X.}, booktitle = {NeurIPS 2020 Workshop on Pre-registration in Machine Learning}, pages = {221--235}, year = {2021}, editor = {Bertinetto, Luca and Henriques, João F. and Albanie, Samuel and Paganini, Michela and Varol, Gül}, volume = {148}, series = {Proceedings of Machine Learning Research}, month = {11 Dec}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v148/bohn21a/bohn21a.pdf}, url = {http://proceedings.mlr.press/v148/bohn21a.html}, abstract = {Humans can learn a variety of concepts and skills incrementally over the course of their lives while exhibiting many desirable properties, such as continual learning without forgetting, forward transfer of knowledge, and learning a new concept with few examples. However, most previous approaches to efficient lifelong learning demonstrate only subsets of these properties, often by different complex mechanisms. In this preregistration submission, we propose to study the effectiveness of a unified lifelong learning framework designed to achieve many of these properties through one central mechanism. We describe this consolidation-based approach and propose experimental protocols to benchmark it on several skills, using grid searches over hyperparameters to better understand the framework.} }
Endnote
%0 Conference Paper %T Towards a Unified Lifelong Learning Framework %A Tanner A. Bohn %A Xinyu Yun %A Charles X. Ling %B NeurIPS 2020 Workshop on Pre-registration in Machine Learning %C Proceedings of Machine Learning Research %D 2021 %E Luca Bertinetto %E João F. Henriques %E Samuel Albanie %E Michela Paganini %E Gül Varol %F pmlr-v148-bohn21a %I PMLR %P 221--235 %U http://proceedings.mlr.press/v148/bohn21a.html %V 148 %X Humans can learn a variety of concepts and skills incrementally over the course of their lives while exhibiting many desirable properties, such as continual learning without forgetting, forward transfer of knowledge, and learning a new concept with few examples. However, most previous approaches to efficient lifelong learning demonstrate only subsets of these properties, often by different complex mechanisms. In this preregistration submission, we propose to study the effectiveness of a unified lifelong learning framework designed to achieve many of these properties through one central mechanism. We describe this consolidation-based approach and propose experimental protocols to benchmark it on several skills, using grid searches over hyperparameters to better understand the framework.
APA
Bohn, T.A., Yun, X. & Ling, C.X.. (2021). Towards a Unified Lifelong Learning Framework. NeurIPS 2020 Workshop on Pre-registration in Machine Learning, in Proceedings of Machine Learning Research 148:221-235 Available from http://proceedings.mlr.press/v148/bohn21a.html.

Related Material