Reward driven discovery of the optimal microstructure representations with invariant variational autoencoders

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Slautin, Boris N., Barakati, Kamyar, Funakubo, Hiroshi, Ziatdinov, Maxim A., Shvartsman, Vladimir V., Lupascu, Doru C., Kalinin, Sergei V.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911187468288000
author Slautin, Boris N.
Barakati, Kamyar
Funakubo, Hiroshi
Ziatdinov, Maxim A.
Shvartsman, Vladimir V.
Lupascu, Doru C.
Kalinin, Sergei V.
author_facet Slautin, Boris N.
Barakati, Kamyar
Funakubo, Hiroshi
Ziatdinov, Maxim A.
Shvartsman, Vladimir V.
Lupascu, Doru C.
Kalinin, Sergei V.
contents Microscopy techniques generate vast amounts of complex image data that in principle can be used to discover simpler, interpretable, and parsimonious forms to reveal the underlying physical structures, such as elementary building blocks in molecular systems or order parameters and phases in crystalline materials. Variational Autoencoders (VAEs) provide a powerful means of constructing such low-dimensional representations, but their performance heavily depends on multiple non-myopic design choices, which are often optimized through trial-and-error and empirical analysis. To enable automated and unbiased optimization of VAE workflows, we investigated reward-based strategies for evaluating latent space representations. Using Piezoresponse Force Microscopy data as a model system, we examined multiple policies and reward functions that can serve as a foundation for automated optimization. Our analysis shows that approximating the latent space with Gaussian Mixture Models (GMM) and Bayesian Gaussian Mixture Models (BGMM) provides a strong basis for constructing reward functions capable of estimating model efficiency and guiding the search for optimal parsimonious representations.
format Preprint
id arxiv_https___arxiv_org_abs_2510_00243
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Reward driven discovery of the optimal microstructure representations with invariant variational autoencoders
Slautin, Boris N.
Barakati, Kamyar
Funakubo, Hiroshi
Ziatdinov, Maxim A.
Shvartsman, Vladimir V.
Lupascu, Doru C.
Kalinin, Sergei V.
Machine Learning
Materials Science
Microscopy techniques generate vast amounts of complex image data that in principle can be used to discover simpler, interpretable, and parsimonious forms to reveal the underlying physical structures, such as elementary building blocks in molecular systems or order parameters and phases in crystalline materials. Variational Autoencoders (VAEs) provide a powerful means of constructing such low-dimensional representations, but their performance heavily depends on multiple non-myopic design choices, which are often optimized through trial-and-error and empirical analysis. To enable automated and unbiased optimization of VAE workflows, we investigated reward-based strategies for evaluating latent space representations. Using Piezoresponse Force Microscopy data as a model system, we examined multiple policies and reward functions that can serve as a foundation for automated optimization. Our analysis shows that approximating the latent space with Gaussian Mixture Models (GMM) and Bayesian Gaussian Mixture Models (BGMM) provides a strong basis for constructing reward functions capable of estimating model efficiency and guiding the search for optimal parsimonious representations.
title Reward driven discovery of the optimal microstructure representations with invariant variational autoencoders
topic Machine Learning
Materials Science
url https://arxiv.org/abs/2510.00243