Summary of Pre-processing and Compression: Understanding Hidden Representation Refinement Across Imaging Domains Via Intrinsic Dimension, by Nicholas Konz et al.
Pre-processing and Compression: Understanding Hidden Representation Refinement Across Imaging Domains via Intrinsic Dimension
by Nicholas Konz, Maciej A. Mazurowski
First submitted to arxiv on: 15 Aug 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG); Image and Video Processing (eess.IV); Machine Learning (stat.ML)
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 paper investigates the changes in intrinsic dimension (ID) of neural networks’ hidden representations across layers, exploring its relationship with generalization ability. The study focuses on domain-specific differences in ID change through layers, analyzing eleven natural and medical image datasets across six network architectures. The findings reveal that medical image models peak in representation ID earlier than natural image models, suggesting differences in abstractness of features used for downstream tasks. A strong correlation between peak representation ID and input space ID is also discovered, indicating that a model’s learned representation information content is guided by its training data. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper looks at how neural networks learn to recognize images from different sources. It tries to understand what makes these networks work well on some types of pictures but not others. The researchers tested 11 groups of image datasets and compared the results with six different network designs. They found that networks trained on medical images changed in a way that was different from those trained on natural images. This change is important for how well the network can recognize things it has never seen before. Overall, the study shows that networks are shaped by what they’re taught, and this has implications for how we use them. |
Keywords
» Artificial intelligence » Generalization
