Summary of Autoencoders in Function Space, by Justin Bunker et al.
Autoencoders in Function Space
by Justin Bunker, Mark Girolami, Hefin Lambley, Andrew M. Stuart, T. J. Sullivan
First submitted to arxiv on: 2 Aug 2024
Categories
- Main: Machine Learning (stat.ML)
- Secondary: 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 The paper introduces function-space versions of deterministic autoencoders (FAEs) and variational autoencoders (FVAEs), which operate directly on functions rather than discretized or pixelated representations. This approach enables the development of better algorithms that smoothly transition between resolutions. The FAE objective is well-defined in many situations, whereas the FVAE objective requires compatibility with the data distribution. Neural operator architectures are paired with these objectives to enable applications such as inpainting, superresolution, and generative modeling of scientific data. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Autoencoders have been widely used for tasks like image processing and scientific applications. Typically, problems are discretized or pixelated before algorithms operate on them. However, this paper shows that considering functions directly can lead to better results. The authors introduce function-space versions of autoencoders (FAEs and FVAEs) and analyze their performance. They also discuss the challenges of defining a well-defined objective for VAEs in function space. |
