Learning what to remember

Robi Bhattacharjee, Gaurav Mahajan
Proceedings of The 33rd International Conference on Algorithmic Learning Theory, PMLR 167:70-89, 2022.

Abstract

We consider a lifelong learning scenario in which a learner faces a neverending and arbitrary stream of facts and has to decide which ones to retain in its limited memory. We introduce a mathematical model based on the online learning framework, in which the learner measures itself against a collection of experts that are also memory-constrained and that reflect different policies for what to remember. Interspersed with the stream of facts are occasional questions, and on each of these the learner incurs a loss if it has not remembered the corresponding fact. Its goal is to do almost as well as the best expert in hindsight, while using roughly the same amount of memory. We identify difficulties with using the multiplicative weights update algorithm in this memory-constrained scenario, and design an alternative scheme whose regret guarantees are close to the best possible.

Cite this Paper

BibTeX
@InProceedings{pmlr-v167-bhattacharjee22a, title = {Learning what to remember}, author = {Bhattacharjee, Robi and Mahajan, Gaurav}, booktitle = {Proceedings of The 33rd International Conference on Algorithmic Learning Theory}, pages = {70--89}, year = {2022}, editor = {Dasgupta, Sanjoy and Haghtalab, Nika}, volume = {167}, series = {Proceedings of Machine Learning Research}, month = {29 Mar--01 Apr}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v167/bhattacharjee22a/bhattacharjee22a.pdf}, url = {https://proceedings.mlr.press/v167/bhattacharjee22a.html}, abstract = {We consider a lifelong learning scenario in which a learner faces a neverending and arbitrary stream of facts and has to decide which ones to retain in its limited memory. We introduce a mathematical model based on the online learning framework, in which the learner measures itself against a collection of experts that are also memory-constrained and that reflect different policies for what to remember. Interspersed with the stream of facts are occasional questions, and on each of these the learner incurs a loss if it has not remembered the corresponding fact. Its goal is to do almost as well as the best expert in hindsight, while using roughly the same amount of memory. We identify difficulties with using the multiplicative weights update algorithm in this memory-constrained scenario, and design an alternative scheme whose regret guarantees are close to the best possible.} }
Endnote
%0 Conference Paper %T Learning what to remember %A Robi Bhattacharjee %A Gaurav Mahajan %B Proceedings of The 33rd International Conference on Algorithmic Learning Theory %C Proceedings of Machine Learning Research %D 2022 %E Sanjoy Dasgupta %E Nika Haghtalab %F pmlr-v167-bhattacharjee22a %I PMLR %P 70--89 %U https://proceedings.mlr.press/v167/bhattacharjee22a.html %V 167 %X We consider a lifelong learning scenario in which a learner faces a neverending and arbitrary stream of facts and has to decide which ones to retain in its limited memory. We introduce a mathematical model based on the online learning framework, in which the learner measures itself against a collection of experts that are also memory-constrained and that reflect different policies for what to remember. Interspersed with the stream of facts are occasional questions, and on each of these the learner incurs a loss if it has not remembered the corresponding fact. Its goal is to do almost as well as the best expert in hindsight, while using roughly the same amount of memory. We identify difficulties with using the multiplicative weights update algorithm in this memory-constrained scenario, and design an alternative scheme whose regret guarantees are close to the best possible.
APA
Bhattacharjee, R. & Mahajan, G.. (2022). Learning what to remember. Proceedings of The 33rd International Conference on Algorithmic Learning Theory, in Proceedings of Machine Learning Research 167:70-89 Available from https://proceedings.mlr.press/v167/bhattacharjee22a.html.

Related Material