Work

Marginal Replay vs Conditional Replay for Continual Learning

Year: 2018

Type: preprint

Abstract: We present a new replay-based method of continual classification learning that we term "conditional replay" which generates samples and labels together by sampling from a distribution conditioned on the class. We compare conditional replay to another replay-based continual learning paradigm (which we term "marginal replay") that generates samples independently of their class and assigns labels in a separate step. The main improvement in conditional replay is that labels for generated samples need not be inferred, which reduces the margin for error in complex continual classification learning tasks. We demonstrate the effectiveness of this approach using novel and standard benchmarks constructed from MNIST and FashionMNIST data, and compare to the regularization-based \textit{elastic weight consolidation} (EWC) method.

Source: arXiv (Cornell University)

DOI: 10.48550/arxiv.1810.12069

Authors: Lesort, Timothée, Gepperth, Alexander, Stoian, Andrei, Filliat, David

Institutions: Institut national de recherche en sciences et technologies du numérique, École Nationale Supérieure de Techniques Avancées, Thales (France), École Nationale Supérieure de Techniques Avancées Paris, Fulda University of Applied Sciences

Language: English

Topic: Domain Adaptation and Few-Shot Learning

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

OpenAlex ID: W4296290295

Cites:

Cited by:

Related to: 10