Inductive Learning Hypothesis
Any hypothesis that can approximate the target function well over a sufficiently large set of training examples can also approximate the target function well over other unobserved examples.
Learning Bias (or Inductive Bias)
This refers to the set of assumptions a learning algorithm makes to predict outputs for new inputs i.e it is “a set of arbitrary assumptions” that enable the learner to generalize. Two main types of bias
- Restriction bias: Limiting the hypothesis space (as in Candidate Elimination)
- Preference bias: Preferring certain hypotheses over others (as in ID3’s greedy search)