具有重复的类增量学习
Class-Incremental Learning with Repetition
AI总结：
针对真实数据流中的概念重复，提出类增量重复场景及两种流生成器，并设计利用重复的重放策略，在CIFAR100和TinyImageNet上优于现有重放方法。
摘要：
真实世界的数据流自然包含先前概念的重复。从持续学习（CL）的角度来看，重复是环境的一种属性，与重放不同，它不能被智能体控制。如今，类增量（CI）场景代表了评估和比较持续学习策略的主要测试平台。这种场景类型非常易于使用，但它从不允许重新访问先前见过的类别，因此完全忽略了重复的作用。我们关注具有重复的类增量（CIR）场景家族，其中重复被嵌入到流的定义中。我们提出了两个随机流生成器，它们从单个数据集和几个可解释的控制参数出发，产生广泛的CIR流。我们通过研究现有持续学习策略在不同CIR流下的行为，进行了首次关于持续学习中重复的全面评估。然后，我们提出了一种新颖的重放策略，该策略利用重复并抵消流中自然存在的不平衡。在CIFAR100和TinyImageNet上，我们的策略优于其他重放方法，这些方法并非为具有重复的环境而设计。
英文摘要：
Real-world data streams naturally include the repetition of previous concepts. From a Continual Learning (CL) perspective, repetition is a property of the environment and, unlike replay, cannot be controlled by the agent. Nowadays, the Class-Incremental (CI) scenario represents the leading test-bed for assessing and comparing CL strategies. This scenario type is very easy to use, but it never allows revisiting previously seen classes, thus completely neglecting the role of repetition. We focus on the family of Class-Incremental with Repetition (CIR) scenario, where repetition is embedded in the definition of the stream. We propose two stochastic stream generators that produce a wide range of CIR streams starting from a single dataset and a few interpretable control parameters. We conduct the first comprehensive evaluation of repetition in CL by studying the behavior of existing CL strategies under different CIR streams. We then present a novel replay strategy that exploits repetition and counteracts the natural imbalance present in the stream. On both CIFAR100 and TinyImageNet, our strategy outperforms other replay approaches, which are not designed for environments with repetition.
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。
