[论文解读] Task-Free Continual Learning via Online Discrepancy Distance Learning

Fei Ye, Adrian G. Borş|arXiv (Cornell University)|Oct 12, 2022

Domain Adaptation and Few-Shot Learning被引用 21

一句话总结

本论文提出 Online Discrepancy Distance Learning (ODDL) 用于 Task-Free Continual Learning (TFCL) 的理论支撑框架，基于 discrepancy distance 的理论，以及动态混合扩展和基于 discrepancy 的记忆采样，在 TFCL 基准上达到最先进的结果。

ABSTRACT

Learning from non-stationary data streams, also called Task-Free Continual Learning (TFCL) remains challenging due to the absence of explicit task information. Although recently some methods have been proposed for TFCL, they lack theoretical guarantees. Moreover, forgetting analysis during TFCL was not studied theoretically before. This paper develops a new theoretical analysis framework which provides generalization bounds based on the discrepancy distance between the visited samples and the entire information made available for training the model. This analysis gives new insights into the forgetting behaviour in classification tasks. Inspired by this theoretical model, we propose a new approach enabled by the dynamic component expansion mechanism for a mixture model, namely the Online Discrepancy Distance Learning (ODDL). ODDL estimates the discrepancy between the probabilistic representation of the current memory buffer and the already accumulated knowledge and uses it as the expansion signal to ensure a compact network architecture with optimal performance. We then propose a new sample selection approach that selectively stores the most relevant samples into the memory buffer through the discrepancy-based measure, further improving the performance. We perform several TFCL experiments with the proposed methodology, which demonstrate that the proposed approach achieves the state of the art performance.

研究动机与目标

• 通过利用 discrepancy distance 和领域自适应概念，为 TFCL 中的遗忘提供理论框架。
• 开发动态扩展机制，通过冻结过往组件来维持紧凑且准确的模型。
• 引入基于 discrepancy 的样本选择策略，以用多样且相关的示例填充记忆。
• 提出一个 ODDL 算法，将基于 VAE 的 discrepancy 估计器与记忆重放记忆缓冲区集成。

提出的方法

• 将领域自适应理论扩展到 TFCL，以基于 discrepancy distance 推导时间相关的泛化风险界限。
• 引入动态扩展模型 (DEM)，在添加新组件的同时冻结先前的组件以减少遗忘。
• 使用基于 discrepancy 的准则来指导何时添加新组件以及如何选择记忆样本。
• 在测试阶段使用 VAE 来估计 discrepancy 并协助组件选择。
• 概述一个端到端的训练过程，包含初始学习、评估器训练和样本选择阶段。

实验结果

研究问题

• RQ1 在没有任务标签的情况下，discrepancy distance 如何量化 TFCL 中的遗忘？
• RQ2 带冻结组件的动态扩展模型能否在 TFCL 下提高泛化能力？
• RQ3 基于 discrepancy 的记忆采样是否提升对既有知识的保留和整体性能？
• RQ4 基于 VAE 的 discrepancy 估计如何促进组件选择与模型扩展？

主要发现

• 该工作提供了一个理论框架，通过 discrepancy distance 为 TFCL 推导时间相关的泛化界限。
• 作者表示，具有动态扩展的 ODDL 在 TFCL 基准测试上达到最先进的结果。
• 基于 discrepancy 的记忆扩展和样本选择策略改善了知识保留和模型紧凑性。
• 冻结组件与新学习组件的混合能够减轻负迁移并促进更好的泛化。
• 该方法将理论保证与一个在没有显式任务边界的 TFCL 下的实用算法相结合。