Scaling Law of Sim2Real Transfer Learning in Expanding Computational Materials Databases for Real-World Predictions
It addresses the challenge of limited experimental data in materials science by providing scaling laws for simulation-to-real transfer learning, which is incremental but practical for researchers developing computational databases.
This study demonstrates that prediction error on real materials systems decreases according to a power-law as computational data size increases, offering insights for database development such as determining necessary sample sizes and guiding data production protocols.
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.