Summary of Real: Representation Enhanced Analytic Learning For Exemplar-free Class-incremental Learning, by Run He et al.
REAL: Representation Enhanced Analytic Learning for Exemplar-free Class-incremental Learning
by Run He, Huiping Zhuang, Di Fang, Yizhu Chen, Kai Tong, Cen Chen
First submitted to arxiv on: 20 Mar 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Computer Vision and Pattern Recognition (cs.CV)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes a novel approach to class-incremental learning (CIL) without available historical data, addressing catastrophic forgetting issues. Building upon recent advancements in analytic learning (AL)-based CIL, the authors introduce Representation Enhanced Analytic Learning (REAL), which enhances the representation of the extractor through dual-stream base pretraining (DS-BPT) and representation enhancing distillation (RED). DS-BPT pretrains the model on both supervised learning and self-supervised contrastive learning (SSCL) for base knowledge extraction, while RED distills supervised knowledge to the SSCL-pretrained backbone, facilitating AL-based CIL that converts CIL to a recursive least-square problem. Experimental results on CIFAR-100, ImageNet-100, and ImageNet-1k demonstrate REAL outperforms state-of-the-art in EFCIL and achieves comparable or superior performance compared to replay-based methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper solves a big problem with machines learning new things. When we want computers to learn from lots of data without forgetting old information, it gets really tricky! The authors create a special way called Representation Enhanced Analytic Learning (REAL) that makes the computer remember old things better and learn new things too. They test REAL on different types of pictures and show that it works much better than other methods! |
Keywords
* Artificial intelligence * Distillation * Pretraining * Self supervised * Supervised
