TL;DR本文提出了基于分类法系列的 Taxonomic Class Incremental Learning (TCIL) 问题,通过参数继承实现了知识的增量传递,实验证明 TCIL 方法在 CIFAR-100 和 ImageNet-100 上均优于 SOTA 方法。
Abstract
The problem of continual learning has attracted rising attention in recent
years. However, few works have questioned the commonly used learning setup,
based on a task curriculum of random class. This differs sign