Inductive bias Machine Learning
noun phrase
Definition: The set of assumptions a learning algorithm uses to generalize from observed examples to unseen inputs [Bakanach 2023/2024].
Example in context: “We assume that some form of structure exists in the problem and/or the solution space, which can be incorporated into the learning pipeline as an inductive bias.” [Mohan et al. 2024]
Related terms: generalization, hypothesis space, prior assumptions