Prompt-Based Exemplar Super-Compression and Regeneration for Class-Incremental Learning
This addresses memory and diversity limitations in class-incremental learning for AI systems, representing a strong incremental advance with specific gains.
The paper tackles the problem of limited exemplar diversity in class-incremental learning by introducing PESCR, which compresses images into prompts to reduce memory usage by 24 times and regenerates diverse exemplars using a diffusion model, achieving a 3.2% improvement over previous state-of-the-art on ImageNet-100.
Replay-based methods in class-incremental learning (CIL) have attained remarkable success. Despite their effectiveness, the inherent memory restriction results in saving a limited number of exemplars with poor diversity. In this paper, we introduce PESCR, a novel approach that substantially increases the quantity and enhances the diversity of exemplars based on a pre-trained general-purpose diffusion model, without fine-tuning it on target datasets or storing it in the memory buffer. Images are compressed into visual and textual prompts, which are saved instead of the original images, decreasing memory consumption by a factor of 24. In subsequent phases, diverse exemplars are regenerated by the diffusion model. We further propose partial compression and diffusion-based data augmentation to minimize the domain gap between generated exemplars and real images. PESCR significantly improves CIL performance across multiple benchmarks, e.g., 3.2% above the previous state-of-the-art on ImageNet-100.