Inductive Bias

Assumptions or preferences built into a machine learning algorithm that help it learn from limited data and generalize to new examples. Without inductive bias, a model would need to see every possible input to make a prediction, which is often not feasible.