A Theoretical Study on Solving Continual Learning
A Theoretical Study on Solving Continual Learning
复制标题
DOI：
10.48550/arxiv.2211.02633
复制
发表时间：
2022-11
期刊：
影响因子：
--
通讯作者：
Gyuhak Kim;Changnan Xiao;Tatsuya Konishi;Zixuan Ke;Bin Liu
中科院分区：
文献类型：
--
作者：
Gyuhak Kim;Changnan Xiao;Tatsuya Konishi;Zixuan Ke;Bin Liu
持续学习（CL）是一系列任务的增量学习。有两种流行的CL设置，类增量学习（CIL）和任务增量学习（TIL）。CL的主要挑战是灾难性遗忘（CF）。虽然已经有许多技术可以有效地克服TIL的CF，但CIL仍然具有高度挑战性。到目前为止，很少有理论研究已经做了一个原则性的指导如何解决CIL问题。本文就是进行这样的研究。它首先表明，从概率上讲，CIL问题可以分解为两个子问题：任务内预测（WP）和任务ID预测（TP）。这进一步证明了TP与分布外（OOD）检测相关，它连接了CIL和OOD检测。本研究的主要结论是，无论WP和TP或OOD检测是否明确或隐含定义的CIL算法，良好的WP和良好的TP或OOD检测是必要的，良好的CIL性能足够。另外，TIL只是WP。在理论分析的基础上，设计了新的CIL方法，在CIL和TIL环境下，新的CIL方法的性能都大大优于强基线。
Continual learning (CL) learns a sequence of tasks incrementally. There are two popular CL settings, class incremental learning (CIL) and task incremental learning (TIL). A major challenge of CL is catastrophic forgetting (CF). While a number of techniques are already available to effectively overcome CF for TIL, CIL remains to be highly challenging. So far, little theoretical study has been done to provide a principled guidance on how to solve the CIL problem. This paper performs such a study. It first shows that probabilistically, the CIL problem can be decomposed into two sub-problems: Within-task Prediction (WP) and Task-id Prediction (TP). It further proves that TP is correlated with out-of-distribution (OOD) detection, which connects CIL and OOD detection. The key conclusion of this study is that regardless of whether WP and TP or OOD detection are defined explicitly or implicitly by a CIL algorithm, good WP and good TP or OOD detection are necessary and sufficient for good CIL performances. Additionally, TIL is simply WP. Based on the theoretical result, new CIL methods are also designed, which outperform strong baselines in both CIL and TIL settings by a large margin.