About ChemAI

ChemAI Platform

• Toolkits
  • 1. Band Gap Prediction
  • 2. Formation Energy Prediction
  • 3. Thermoelectricity Prediction
  • 4. Band Gap Prediction of Perovskites
  • 5. Band Gap Correction
  • 6. Prediction of Absorption Max
  • 6. Data Clustering
  • 7. Outlier Detection
• Machnie Learning
  • Model Training
    • 1. Training Dataset Upload (Step 1)
      • 1.1. Dataset Formats
        • 1.1.1. Chemical Formula
        • 1.1.2. SMILES
        • 1.1.3. Crystal Structure
        • 1.1.4. Numeric Vector
      • 1.2. Problem Definition
    • 2. Model Configuration (Step 2)
      • 2.1. Model Selection
      • 2.2. Hyperparameter setting
    • 3. Training Options (Step 3)
      • 3.1. Training configuration
      • 3.2. Outlier Removal
      • 3.3. Transfer learning
    • 4. Training Job (Step 4)
  • Training Results
    • 1. Evaluation Results
      • 1.1. Prediction results
      • 1.2. Target distribution
      • 1.3. Error distribution
    • 2. Actions
      • 2.1. Download model
      • 2.2. Download interpretation
      • 2.3. Store model
    • 3. Training configurations

1. Goal of ChemAI

In the field of chemical material, various kinds of chemical (or material) data are poured out every day. Numerous data containing important information from the chemical composition, synthesis method, analysis conditions to analysis results for certain materials are being produced even at this moment and are being stored somewhere on the researcher’s local computer. Some data is stored somewhere on the individual researcher’s local computer, so you don’t even know if the data is there or not, while others are collected by arranging and collecting the experimental results on a server through a Raspberry Pi in real-time. When high-quality data begins to gather, the potential to utilize data begins to emerge in proportion to the amount of data. The more high-quality data is collected, the more the potential for data utilization increases in proportion to the amount of data.

ChemAI’s first goal

• To maximize the utilization of accumulated chemical data.

ChemAI’s first goal is to maximize the utilization of accumulated chemical data. This includes all the processes of building a predictive model for physical properties desired by a user based on accumulated data, finding input conditions with optimal properties based on the prediction model, or using the prediction model built by other users. Furthermore, when predicting physical properties through machine learning or performing inverse material engineering using predictive models, the user should prepare and carry out various things for that purpose. It is necessary to prepare the dataset, study and prepare the general matters for performing machine learning and prepare the necessary hardware and software. This series of preparation courses can be the barriers to entry for AI research.

ChemAI’s second goal

• To lower the barriers to entry for AI research.

ChemAI ‘s second goal is to lower the height of this barrier. Even if the knowledge and interest in artificial intelligence are small, it is to make it easy for anyone to access the world of artificial intelligence. ChemAI supports to directly generate the predicted results of arbitrary materials and utilizing the predicted results to be used in individual research areas of users.

2. What is Machine Learning?

The whole concept of machine learning is figuring out ways in which we can teach a computer to perform a task without needing to provide explicit instructions. Another way to think about it is that we’re trying to “program” intuition in a computer. You and I can look at an email and easily discern whether or not it’s spam, but how do you get a computer to do such a task? You could construct a huge convoluted logic infrastructure of “if.. then..” statements to sort out the spam emails, but it would be a pain to construct and probably wouldn’t work too well. Instead, the machine learning approach is to equip the computer with skills to learn on its own and feed it a bunch of examples. Machine learning is exploding as a field right now as people are realizing a multitude of tasks that we can teach computers to perform by feeding it large datasets.

3. Overview

The ChemAI provides four main functions: (1) the Toolkits that allow the user to directly predict various physical properties and lets the user experience the allure of artificial intelligence, (2) the Model training that supports users to build prediction models using the user-owned dataset, (3) the Applications that allows you to share your prediction models with other users, and (4) the Datasets that provides example datasets and their information for users who do not have own dataset.

Indices and tables

• Index

• Module Index

• Search Page