Summary of Material Microstructure Design Using Vae-regression with Multimodal Prior, by Avadhut Sardeshmukh et al.
Material Microstructure Design Using VAE-Regression with Multimodal Prior
by Avadhut Sardeshmukh, Sreedhar Reddy, BP Gautham, Pushpak Bhattacharyya
First submitted to arxiv on: 27 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Materials Science (cond-mat.mtrl-sci); 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 Our research proposes a novel approach to building structure-property linkages in computational materials science using variational autoencoders (VAEs) and regression techniques. The method combines the strengths of both models by conditioning a two-level prior on regression variables, optimizing joint losses for reconstruction and prediction. This enables forward prediction of properties given microstructures and inverse prediction of microstructures required to achieve specific properties. To address the ill-posed nature of the inverse problem, we derive an objective function using a multi-modal Gaussian mixture prior, allowing for inference of multiple microstructures per target property set. Our results show that our method is comparable in accuracy to state-of-the-art forward-only models and enables direct inverse inference, eliminating the need for optimization loops. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary In simple terms, this research creates a new way to connect the structure of materials with their properties using machine learning tools. It combines two existing methods to make predictions about what a material will do based on its internal structure, as well as predicting what kind of structure is needed to achieve certain properties. This helps solve a long-standing problem in computer science and materials engineering. |
Keywords
* Artificial intelligence * Inference * Machine learning * Multi modal * Objective function * Optimization * Regression
