Future-Proofing Class-Incremental Learning
This addresses the challenge of data dependency in incremental learning for AI systems, offering a less costly pre-training method, though it is incremental as it builds on existing feature extractor approaches.
The paper tackles the problem of exemplar-free class incremental learning, where replay memory is unavailable, by using a pre-trained text-to-image diffusion model to generate synthetic images of future classes to train feature extractors, improving state-of-the-art methods on benchmarks like CIFAR100 and ImageNet-Subset, especially in settings with few classes in the first incremental step.
Exemplar-Free Class Incremental Learning is a highly challenging setting where replay memory is unavailable. Methods relying on frozen feature extractors have drawn attention recently in this setting due to their impressive performances and lower computational costs. However, those methods are highly dependent on the data used to train the feature extractor and may struggle when an insufficient amount of classes are available during the first incremental step. To overcome this limitation, we propose to use a pre-trained text-to-image diffusion model in order to generate synthetic images of future classes and use them to train the feature extractor. Experiments on the standard benchmarks CIFAR100 and ImageNet-Subset demonstrate that our proposed method can be used to improve state-of-the-art methods for exemplar-free class incremental learning, especially in the most difficult settings where the first incremental step only contains few classes. Moreover, we show that using synthetic samples of future classes achieves higher performance than using real data from different classes, paving the way for better and less costly pre-training methods for incremental learning.