Skip to content

research(model): learned mixture model — suppression & synergy (data-gated) #252

Description

@rvnminers-A-and-N

Today a blend is read component-wise: each molecule's profile is predicted independently and combined arithmetically (see mixture_to_molecule). Real mixtures do not work that way — components suppress and enhance each other, which is most of the craft in formulation.

This needs panel data on MIXTURES (perceived profile of a blend vs its components), which no public dataset provides — hence data-gated, like odour intensity. Worth stating the requirement precisely so it can be scoped if a customer brings their own panel data, and so the current arithmetic combination is not mistaken for a mixture model.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Labels

area:trainingPython dataset build + model trainingenhancementNew feature or request

Type

No type

Projects

No projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions