Home / Product / What is the Citrine Platform? / Artificial Intelligence

Artificial Intelligence

Cutting-edge machine learning techniques, tailored specifically for materials and chemicals

What is it used for?

Optimize composition
and processing to hit
new performance specifications

Panasonic organic semiconductor

Machine Learning Accelerates Materials Development

Learn how the Citrine Platform helped Panasonic develop new organic semiconductors with 25% higher hole mobility.

Quickly reformulate to reduce recipe costs AND improve supply chain resiliency

New AI generated recipes reduce costs

Reduced Costs and Increased Supply Resilience

Learn how Citrine’s AI Platform can be used to reduce costs and increase supply chain resilience in specialty chemicals.

Make
project-level
R&D
decisions

glass

Data-Driven Research Strategy

Learn how data-driven assessments of research directions drive R&D strategy at a global glass manufacturer.

Citrine's Approach to AI

domain knowledge integration into AI models

An easily interpretable graphical model

WORKING WITH SMALL DATA SETS

The Citrine Platform uses algorithms that work well on the small, sparse data sets usually available in materials companies. Citrine has tackled projects with fewer than 30 initial data points!

LEVERAGING DOMAIN KNOWLEDGE

A graphical model enables researchers to see how components fit together and facilitates the integration of expert domain knowledge. In the example here, a known relationship between binding energy and selectivity is leveraged. The machine learning models are used to fill in gaps in knowledge, rather than to relearn the basics.

Sequential learning for artificial intelligence

REUSABLE COMPONENTS

Once built, each component can be used as part of other machine learning workflows, contributing to a library of codified domain knowledge that researchers across the team can use.

SEQUENTIAL LEARNING WORKFLOWS

The next generation of design of experiments is facilitated by the Citrine Platform. The iterative processes of Sequential Learning are coupled with systematic uncertainty calculations to enable efficient, systematic exploration of parameter spaces.