> **inductive**: characterized by the inference of [[abstract|general]] laws from particular instances aka "learning" (from [[dataset|data]]) | [[induct\|inductive]] bias / "learning" | component of [[developmental psychology]] / [[human inductive bias]] | component of [[optimization\|machine learning method]] | [[machine inductive bias]] | | ---------------------------------------------------- | -------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------ | --------------------------- | | constraints on [[function space]] / hypothesis class | [[innate]] / [[gene]]tic | [[statistical model]] ([[unsupervised]]) / [[prediction rule\|prediction rule class]] ([[supervised]]) | [[architectural bias]] | | [[optimization]] / learning [[algorithm]] | [[human cognitive skill]] | see [[optimization algorithm]] | [[implicit regularization]] | | objective | [[functionalism]] / [[culture]] | [[loss function]] | [[regularization term]] | ^inductive-bias ![[falsify]] # sources - [[2019HastieEtAlSurprisesHighdimensionalRidgeless|Surprises in High-Dimensional Ridgeless Least Squares Interpolation]] - [[2022SoudryEtAlImplicitBiasGradient|gradient descent on solvable logistic regression converges to the SVM direction]] - [[2021GrosseCSC2541Winter2021]] lec 1 - [[neural kernel]] - [[2022-10-21 edge of stability representations reading group]] - [[STAT 195]] 2022-03 - [[human memory]] - [[machine learning]]