元学习持续学习算法
Metalearning Continual Learning Algorithms
AI总结：
本文提出 Automated Continual Learning（ACL），通过训练自指神经网络元学习自身的上下文内持续学习算法，在无回放 Split-MNIST 等设置中缓解灾难性遗忘并优于多种基线方法。
摘要：
通用学习系统应当在不断变化的环境中以开放式的方式改进自身。然而，传统的神经网络学习算法会遭受灾难性遗忘（CF），即在学习新任务时会遗忘先前获得的技能。我们没有手工设计避免 CF 的新算法，而是提出 Automated Continual Learning（ACL），用于训练自指神经网络，使其元学习属于自身的上下文内持续（元）学习算法。ACL 将持续学习（CL）的理想目标——在旧任务和新任务上都具有良好性能——编码到其元学习目标中。我们的实验表明，ACL 有效解决了朴素上下文内学习算法所面临的“上下文内灾难性遗忘”问题；在无回放设置下的 Split-MNIST 基准测试中，ACL 学到的算法优于手工设计的学习算法和流行的元持续学习方法，并且能够持续学习由多个标准图像分类数据集组成的多样化任务。我们还通过将 ACL 与利用预训练模型的最先进 CL 方法进行比较，讨论了当前上下文内 CL 的局限性。总体而言，我们为长期存在的 CL 问题带来了若干新的视角。
英文摘要：
General-purpose learning systems should improve themselves in open-ended fashion in ever-changing environments. Conventional learning algorithms for neural networks, however, suffer from catastrophic forgetting (CF), i.e., previously acquired skills are forgotten when a new task is learned. Instead of hand-crafting new algorithms for avoiding CF, we propose Automated Continual Learning (ACL) to train self-referential neural networks to metalearn their own in-context continual (meta)learning algorithms. ACL encodes continual learning (CL) desiderata -- good performance on both old and new tasks -- into its metalearning objectives. Our experiments demonstrate that ACL effectively resolves "in-context catastrophic forgetting," a problem that naive in-context learning algorithms suffer from; ACL-learned algorithms outperform both hand-crafted learning algorithms and popular meta-continual learning methods on the Split-MNIST benchmark in the replay-free setting, and enables continual learning of diverse tasks consisting of multiple standard image classification datasets. We also discuss the current limitations of in-context CL by comparing ACL with state-of-the-art CL methods that leverage pre-trained models. Overall, we bring several novel perspectives into the long-standing problem of CL.
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。
