Using machine learning to explore formulations recipes with new ingredients

Abstract

A system and a method are disclosed that, in an embodiment, receive first input from a user of a candidate formulation recipe, and second input from the user of target properties and target property constraints. The system inputs the first input into a machine learning model, the model having been trained using historical training data, each element of the historical training data corresponding to a known formulation having a known feature representation, each known formulation having associated properties and statistical representations of each feature of the known formulation that form the known feature representation. The system receives as output from the model a predicted property of a candidate formulation derived using the first input and the likelihood that the candidate formulation satisfies the target property constraints using the second input. The system generates for display to the user a predicted likelihood that the predicted property satisfies the second input.