Jul 2, 2013 — Elastic Net: Flexible Regularization for Linear Regression — As a reminder, a regularization technique applied to linear regression helps us to select the most relevant features, x, to predict an outcome y.  For now, see my post about LASSO for more details about regularization.  Both LASSO and elastic net,... Read more

Jul 2, 2013 — LASSO: Regularization for Linear Regression — From the mind of the master, we can define lasso as follows: “The Lasso is a shrinkage and selection method for linear regression. It minimizes the usual sum of squared errors, with a bound on the sum of the absolute... Read more

Jul 1, 2013 — Expectation Maximization (EM) Algorithm — Let’s talk about jelly beans.  Specifically, imagine that you took a bag of every single brand of jelly bean in the world (meaning different colors, sizes, and ingredients) and dumped them into a bin.  Your bin now has some k... Read more

Jun 30, 2013 — Region Growing — Region Growing is an algorithm for segmentation of regions in an image based on the idea that regions share some property that can be computationally measured.  It’s one of those magical algorithms that is ridiculously simply, and powerful because the... Read more