arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2006.04027cs.LGcs.AIstat.ML
收藏

面向持续学习的高效体系结构搜索

Efficient Architecture Search for Continual Learning

University of Electronic Science and Technology of China(电子科技大学)

Qiang Gao, Zhipeng Luo, Diego Klabjan

摘要:

使用神经网络的持续学习是AI中的一个重要学习框架,旨在很好地学习一系列任务。然而,它经常面临三个挑战:(1)克服灾难性遗忘问题,(2)使当前网络适应新的任务,同时(3)控制其模型复杂性。为了达到这些目标,我们提出了一种新的方法命名为持续学习与有效的架构搜索,或简称CLEAS。CLEAS与神经架构搜索(NAS)密切合作,NAS利用强化学习技术搜索适合新任务的最佳神经架构。特别是,我们设计了一个神经元级的NAS控制器,它决定了以前任务中的哪些旧神经元应该被重用(知识转移),以及哪些新神经元应该被添加(学习新知识)。这样一个细粒度的控制器允许人们找到一个非常简洁的架构,可以很好地适应每个新任务。同时,由于我们没有改变重用神经元的权重,我们完美地记住了从以前的任务中学到的知识。我们在许多顺序分类任务上评估了CLEAS,结果表明CLEAS优于其他最先进的替代方法,在使用更简单的神经架构的同时实现了更高的分类精度。

英文摘要:

Continual learning with neural networks is an important learning framework in AI that aims to learn a sequence of tasks well. However, it is often confronted with three challenges: (1) overcome the catastrophic forgetting problem, (2) adapt the current network to new tasks, and meanwhile (3) control its model complexity. To reach these goals, we propose a novel approach named as Continual Learning with Efficient Architecture Search, or CLEAS in short. CLEAS works closely with neural architecture search (NAS) which leverages reinforcement learning techniques to search for the best neural architecture that fits a new task. In particular, we design a neuron-level NAS controller that decides which old neurons from previous tasks should be reused (knowledge transfer), and which new neurons should be added (to learn new knowledge). Such a fine-grained controller allows one to find a very concise architecture that can fit each new task well. Meanwhile, since we do not alter the weights of the reused neurons, we perfectly memorize the knowledge learned from previous tasks. We evaluate CLEAS on numerous sequential classification tasks, and the results demonstrate that CLEAS outperforms other state-of-the-art alternative methods, achieving higher classification accuracy while using simpler neural architectures.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑