Iori Kurata

Title: Iori Kurata: Patent Work in Machine-Learning-Based Materials Research.

Introduction

Iori Kurata is an inventor based in Tokyo, Japan, and works at Preferred Networks, Inc. His patent work includes a training approach for models that estimate the energy of atomic structures involving adsorbed molecules.

Latest Patents

Kurata is associated with 1 patent: “Training Device, Method, Non-Transitory Computer Readable Medium, and Inferring Device.” The invention describes a processor that inputs atomic structures into a model and compares the model’s predicted energies with ground-truth values. It uses errors from structures containing a surface and structures containing a cluster, each with an adsorbed molecule nearby, to update a model parameter. The surface and cluster contain the same atom.

Career Highlights

Kurata’s patent reflects work at the intersection of machine learning and atomic-scale modeling. The described training method uses examples from both surface and cluster structures to improve energy predictions.

Collaborations

Kurata’s named coworkers include Chikashi Shinagawa and So Takamoto. His work is associated with Preferred Networks, Inc.

Conclusion

Iori Kurata’s patent documents a model-training method for predicting energies in atomic structures, highlighting his contribution to machine-learning-based materials research.