All of Machine Learning or certainly supervised learning is about induction

1.Induction and Deduction

To make a good machine learning we must collect a good data, understand your data and know how to manipulate them, also you must have a technique to explore data, understand it, to be capable to integrate algorithms,…And if you don’t master the technique you will never be a great machine learning persons, and you will get a bad machine learning.

After collecting good data, the biggest fundamental problem on machine learning is how to generalize a function who is uniform with those data. All of machine learning or certainly supervised learning is about induction, as opposed to deduction.

1.2.Induction

In Machine Learning, induction is not only taken as the inference from observations to given general rules. It includes the search for these rules in a large set of possibilities (refernce : from F. Bergadano* University of Torino Corso Svizzera 1S5, Torino, Italy bergadan@di.unito.it ). Induction being the problem of going from specific examples to a more general rule or conclusion.We talk about bias and in particular inductive bias. We can properly say that machine learning is the induction of an approximate function (general function) based on a good data.

1.2.Deduction

It is the fact of going from a general rule to specific instances. Many AI are based on deductive reasoning and logics, you deduce answers from exinsting rules. For example you have a rule in universe that tell you that "All men are mortal" (general) and "Socrates is a man" (specific) so we deduct that Socrates is mortal.