Summary of An Attempt to Generate New Bridge Types From Latent Space Of Generative Flow, by Hongjun Zhang
An attempt to generate new bridge types from latent space of generative flow
by Hongjun Zhang
First submitted to arxiv on: 18 Jan 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); 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 A novel approach to normalizing flows is introduced through simple examples and concise explanations. The paper explores the essence of probability transformation from the perspective of the distribution of random variables, introducing the scaling factor Jacobian determinant. The authors demonstrate how normalizing flows cleverly solve high-dimensional matrix determinant calculation and neural network reversible transformation challenges. Using a structured image dataset of bridges, the authors construct and train a normalizing flow model based on the Glow API in TensorFlow Probability library. The model transforms the complex bridge distribution into a standard normal distribution, enabling the generation of new bridge types different from the training dataset. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Normalizing flows are a way to change the shape of data distributions. Imagine you have a set of random variables that follow certain patterns. This paper explains how to transform those patterns into simpler ones using math and computer algorithms. The authors show that this can help solve two big problems in machine learning: calculating large matrices and making neural networks reversible. They use a dataset of pictures of different types of bridges to demonstrate their approach, which can generate new bridge designs that are not seen before. |
Keywords
* Artificial intelligence * Machine learning * Neural network * Probability
