•Under-fitting:
–High bias (models are not accurate)
–Small variance
(smaller influence
of examples in the training
set)
•Over-fitting:
–Small bias (models flexible enough to fit well to training data)
–Large variance
(models depend
very much on the training
set)