AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Chemo Informatics - Is There a Chemistry Aware Validation Process?
In the real world, machine learning often doesn't generalize. You're most interested in predicting the structure of molecules that you haven't seen. So what you want to do is when doing train valid tests, you often want to do a splitting of the data where the validation and test sets are drawn from chemical scaffolds. And this gives you a better estimation of the true generalizability of the model.