Saved in:
Bibliographic Details
Main Authors: Minami, Shunya, Hayashi, Yoshihiro, Wu, Stephen, Fukumizu, Kenji, Sugisawa, Hiroki, Ishii, Masashi, Kuwajima, Isao, Shiratori, Kazuya, Yoshida, Ryo
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2408.04042
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909281450721280
author Minami, Shunya
Hayashi, Yoshihiro
Wu, Stephen
Fukumizu, Kenji
Sugisawa, Hiroki
Ishii, Masashi
Kuwajima, Isao
Shiratori, Kazuya
Yoshida, Ryo
author_facet Minami, Shunya
Hayashi, Yoshihiro
Wu, Stephen
Fukumizu, Kenji
Sugisawa, Hiroki
Ishii, Masashi
Kuwajima, Isao
Shiratori, Kazuya
Yoshida, Ryo
contents To address the challenge of limited experimental materials data, extensive physical property databases are being developed based on high-throughput computational experiments, such as molecular dynamics simulations. Previous studies have shown that fine-tuning a predictor pretrained on a computational database to a real system can result in models with outstanding generalization capabilities compared to learning from scratch. This study demonstrates the scaling law of simulation-to-real (Sim2Real) transfer learning for several machine learning tasks in materials science. Case studies of three prediction tasks for polymers and inorganic materials reveal that the prediction error on real systems decreases according to a power-law as the size of the computational data increases. Observing the scaling behavior offers various insights for database development, such as determining the sample size necessary to achieve a desired performance, identifying equivalent sample sizes for physical and computational experiments, and guiding the design of data production protocols for downstream real-world tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2408_04042
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Scaling Law of Sim2Real Transfer Learning in Expanding Computational Materials Databases for Real-World Predictions
Minami, Shunya
Hayashi, Yoshihiro
Wu, Stephen
Fukumizu, Kenji
Sugisawa, Hiroki
Ishii, Masashi
Kuwajima, Isao
Shiratori, Kazuya
Yoshida, Ryo
Materials Science
Machine Learning
To address the challenge of limited experimental materials data, extensive physical property databases are being developed based on high-throughput computational experiments, such as molecular dynamics simulations. Previous studies have shown that fine-tuning a predictor pretrained on a computational database to a real system can result in models with outstanding generalization capabilities compared to learning from scratch. This study demonstrates the scaling law of simulation-to-real (Sim2Real) transfer learning for several machine learning tasks in materials science. Case studies of three prediction tasks for polymers and inorganic materials reveal that the prediction error on real systems decreases according to a power-law as the size of the computational data increases. Observing the scaling behavior offers various insights for database development, such as determining the sample size necessary to achieve a desired performance, identifying equivalent sample sizes for physical and computational experiments, and guiding the design of data production protocols for downstream real-world tasks.
title Scaling Law of Sim2Real Transfer Learning in Expanding Computational Materials Databases for Real-World Predictions
topic Materials Science
Machine Learning
url https://arxiv.org/abs/2408.04042