Deduction
It is a way of arriving at a type of reasoning that by using multiple theorems as a basis. Example: No fruit is savoury, apple is a fruit. Therefore Apple is not savoury.
Abduction
You apply a logic/rule/commonsense to an observation and derive a plausibility. Example: Small chunk of an apple is missing. Therefore a person must have taken a bite.
Induction
You use data to derive a rule/theory that is applicable more generally. Example: A whole number between 1-100 when multiplied with 2 gives an even number. Therefore every whole number multiplied by two is also an even number.
Note Machine-Learning is inductive
Empiricism
It is a valid scientific means to generate knowledge if we accept that hypotheses case only be tested and falsified, but never proven. (More: K. Popper) (Co-relation is not causation)
Inductive Bias
The inductive bias of a learning algorithm is a set of assumptions that a learner uses to generalize for situations that are beyond or not present in the training data.