Fig. 1: Abstract Representation of a funnel based screening approach to molecular discovery.
From: A multi-fidelity machine learning approach to high throughput materials screening
An entire pool of molecules is evaluated with the cheapest and assumed least accurate method, a fixed fraction of the highest performing species are then evaluated with the next cheapest and assumed second least accurate method passing on a fixed fraction of the highest performing species to the next method. In this way the pool of candidate molecules progressively reduced until a final set of candidates remain.
