Supervised Learning

Question 1

What is supervised learning?

Answer 1

Learns a mapping from inputs x = (x1,…,xd)^T ∈ X to outputs y ∈ Y given a training set of input-output pairs T = {(x1, y1),…,(xn, yn)} where d = dimensions and n = size of training set.

Question 2

Can dimensions be different types?

Answer 2

Yes, dimensions can be of different times. Many assume numeric inputs but some accept ordinal, categorical or non-structured data types.

Question 3

What is the hypothesis set?

Answer 3

The hypothesis set is the set of all possible functions that a machine learning approach can learn.

Question 4

What is the aim of the learning algorithm?

Answer 4

The learning algorithm searches for the hypothesis g ∈ H that best appropriates f given T

Question 5

What is the hypothesis formula?

Answer 5

h(x) = wTx + b

Question 6

How do we account for noise?

Answer 6

By considering a distribution over the target function, input, final hypothesis and target.

Question 7

What is the unknown target function distribution?

Answer 7

f : p(x|y)

Question 8

What is the unknown input distribution?

Answer 8

p(x)

Question 9

What is the final hypothesis distribution?

Answer 9

p(y|x)

Question 10

What is the unknown target distribution?

Answer 10

p(y|x)