Optimize mixtures-of-mixtures using artificial intelligence
This white paper discusses how the Alchemite™ software solves a problem where alternative machine learning approaches often fail to find the best solution – how to optimize formulations consisting of components that are themselves mixtures.
Executive Summary
Machine learning delivers outstanding benefits in the design and analysis of formulations, chemicals, materials, biopharmaceuticals, and beyond. Often the components for these products are themselves mixtures and it is the relative amounts of these underlying ingredients that drives the final product’s performance. Yet many experimental design tools do not adequately account for this ‘mixture-of-mixtures’ scenario. In this white paper, we demonstrate how Alchemite™ machine learning offers a workflow that allows designers to specify the available feedstocks, then seamlessly optimize feedstock levels, all whilst retaining the scientific knowledge about the underlying ingredients. This enables users to design practical formulations with confidence that they will perform.
