January 2019
Intermediate to advanced
390 pages
9h 16m
English
With so many models, one always wonders which one to use. Wolpert, in his famous paper The Lack of A Priori Distinctions Between Learning, explored this issue and showed that if we make no prior assumption about the input data, then there's no reason to prefer one model over any other. This is known as the No Free Lunch theorem.
This means that there's no model hat can be a priori guaranteed to work better. The only way we can ascertain which model is best is by evaluating them all. But, practically, it isn't possible to evaluate all of the models and so, in practice, we make reasonable assumptions about the data and evaluate a few relevant models.
Read now
Unlock full access