BriefGPT.xyz
Aug, 2022
连续学习中重放策略的基准和实证分析
A Benchmark and Empirical Analysis for Replay Strategies in Continual Learning
HTML
PDF
Qihan Yang, Fan Feng, Rosa Chan
TL;DR
本篇文章探讨了如何通过回放记忆的方式来解决深度神经网络在连续学习时的“灾难性遗忘”问题，并对各种取样策略下回放数据的效率、性能和可伸缩性进行了深入评估，最终提供了一个针对各种数据分布选择回放方法的实用解决方案。
Abstract
With the capacity of
continual learning
, humans can continuously acquire knowledge throughout their lifespan. However,
computational systems
are not, in general, capable of learning tasks sequentially. This long-
→