From Bioinformatics to AI: 14. Regression, Bias-Variance Tradeoff, and Regularization

Synopsis: Is machine learning equivalent to least squares fitting? If it is more than fitting data to parametrized functions, what principle is out there to define the "optimal" fitting and what practice is out there to reach such an optimal scenario?

Keywords: Regression, Expected Generalization Error, Bias-Variance Tradeoff, Model Complexity, Regularization, Shrinkage Methods, Ridge Regression, LASSO Regression, Tuning Regularization Hyperparameters

Course "Algorithms in Structural Bioinformatics" (Spring 2022)
ALG STRUCT BIOINFORMATICS 3 3 2022