Summary of Explora: Parameter-efficient Extended Pre-training to Adapt Vision Transformers Under Domain Shifts, by Samar Khanna et al.
ExPLoRA: Parameter-Efficient Extended Pre-Training to Adapt Vision Transformers under Domain Shifts
by Samar Khanna, Medhanie Irgau, David B. Lobell, Stefano Ermon
First submitted to arxiv on: 16 Jun 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
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 The authors introduce ExPLoRA, a parameter-efficient fine-tuning technique that adapts pre-trained vision transformers (ViTs) to new domains via self-supervised pre-training. By initializing ViT with pre-trained weights and continuing unsupervised pre-training on the new domain, unfreezing 1-2 blocks and tuning all other layers with LoRA, they achieve state-of-the-art results on satellite imagery tasks while using fewer parameters than fully pre-trained models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper shows how to make computers better at recognizing things in pictures. They use a special way of training the computer called “ExPLoRA” that helps it learn from new types of pictures without needing lots of labeled examples. This is important because it means we can teach computers to do tasks like recognizing objects in satellite images, even if those images are very different from what they were trained on. |
Keywords
» Artificial intelligence » Fine tuning » Lora » Parameter efficient » Self supervised » Unsupervised » Vit
