Summary of A Hard-to-beat Baseline For Training-free Clip-based Adaptation, by Zhengbo Wang et al.
A Hard-to-Beat Baseline for Training-free CLIP-based Adaptation
by Zhengbo Wang, Jian Liang, Lijun Sheng, Ran He, Zilei Wang, Tieniu Tan
First submitted to arxiv on: 6 Feb 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
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 A novel approach to fine-tuning Contrastive Language-Image Pretraining (CLIP) is proposed, which leverages Gaussian Discriminant Analysis (GDA) to enhance its performance in downstream tasks. By applying GDA to the classification of CLIP, the method can estimate class means and covariance without requiring additional training time or computational resources. The ensemble approach combines the original zero-shot classifier with the GDA-based classifier, demonstrating superior results on 17 datasets for few-shot classification, imbalanced learning, out-of-distribution generalization, base-to-new generalization, and unsupervised learning. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A new way to make computers better at understanding pictures and words is presented. This method uses an old idea called Gaussian Discriminant Analysis (GDA) to help a popular AI tool called Contrastive Language-Image Pretraining (CLIP). The result is that CLIP can learn faster and use less computer power. The method works well on many different tasks, like recognizing objects in pictures or understanding spoken language. |
Keywords
* Artificial intelligence * Classification * Few shot * Fine tuning * Generalization * Pretraining * Unsupervised * Zero shot
