**Consider
**

- concept learning algorithm
*L* - instances
*X*, target concept*c* - training examples
- let
*L*(*x*_{i},*D*_{c}) denote the classification assigned to the instance*x*_{i}by*L*after training on data*D*_{c}.

**
**

**Formal Definition**:

Theinductive biasofLis any minimal set of assertionsBsuch that for any target conceptcand corresponding training examplesD_{c}

**where **
** means ***A*** logically entails ***B*