CrowdChem, Inc.
Development of a materials platform for integrating internal and external data in the materials industry

Program name

Deep-Tech Startups Support Program

Business Plan

We build a database by manually extracting information such as compounding information and experimental results from examples of chemical product structures and properties, mainly from publicly available web catalogs, and chemical experts from examples of chemical patents. By using a proprietary machine learning model to learn from the data set obtained in this extraction process, it is possible to predict the resulting property information from compounding information and processing processes, and to search for combinations of raw materials that will achieve the desired properties. Based on this model, the company provides compounding analysis and proposal services and platform services.

Research Outline

By developing federated learning technology based on a unique machine learning algorithm that can connect data resources across the materials industry based on catalogs and patents and different data, we will commercialize a platform where all companies that manufacture and use materials can share and use each other's data.

1. Development of federated learning technology for the chemical industry
2. Improvement of the accuracy of federated learning models
3. Development of information security technology for practical use of federated learning

International collaborative technology demonstration

• Download this page（198KB）