Continual Learning in Linear Classification on Separable Data

Authors: Itay Evron, Edward Moroshko, Gon Buzaglo, Maroun Khriesh, Badea Marjieh, Nathan Srebro, Daniel Soudry

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We theoretically study the continual learning of a linear classification model on separable data with binary classes. Even though this is a fundamental setup to consider, there are still very few analytic results on it, since most of the continual learning theory thus far has focused on regression settings.
Researcher Affiliation Academia 1Department of Electrical and Computer Engineering, Technion, Haifa, Israel 2Toyota Technological Institute at Chicago, Chicago IL, USA.
Pseudocode Yes Scheme 1 Regularized Continual Learning Initialization: w(λ) 0 = 0D Iterative update for each task t [k]: w(λ) t = argmin w RD (x,y) St e yw x + λt w w(λ) t 1 2
Open Source Code No The paper does not provide an explicit statement or link for open-source code availability.
Open Datasets No The paper refers to 'separable datasets' and 'task sequences' but does not specify or provide access information for any publicly available or open datasets used in its analysis or illustrations.
Dataset Splits No The paper focuses on theoretical analysis and does not describe empirical experiments requiring training/test/validation dataset splits.
Hardware Specification No The paper is theoretical and does not mention specific hardware used for any computational work or simulations.
Software Dependencies No The paper does not mention specific software dependencies with version numbers.
Experiment Setup No The paper defines theoretical schemes and parameters (e.g., λ, p) for its analysis, but does not provide concrete experimental setup details like hyperparameters or training configurations for empirical validation.