| A learning algorithm is a function
which maps each attribute vector
target value .
| The empirical error on the training
set is always lower than the empirical error on the test set.
| The true error of the hypothesis is
necessarily larger then the empirical error
measured on the test set
| If the training set
and the test set are generated by two totally
different distributions, then
| Generalization has to do
| A ``good'' learning algorithm is a
learning algorithm which