Aug 13, 2013 — Kernel Methods — Kernel methods, most commonly described in relation to support vector machines, allow us to transform our data into a higher dimensional space to assist with the goal of the learning algorithm.  Specifically for SVMs, we are looking to define the... Read more

Aug 9, 2013 — The Gap Statistic — The gap statistic is a method for approximating the “correct” number of clusters, k, for an unsupervised clustering. We do this by assessing a metric of error (the within cluster sum of squares) with regard to our choice of k.... Read more

Aug 6, 2013 — Segmentation of MRI with Bayesian Methods — Let’s say that I am a proponent of the idea that I can segment the brain into meaningful regions (e.g., caudate, amygdala, hippocampus…), which some may think is akin to cutting up the United States into said states.  This could... Read more

Aug 4, 2013 — Dimensionality Estimation of ICA with Bayesian Analysis — Across methods of matrix decomposition (ICA, Archetypal Analysis, NNMF), we run into the problem of needing to know how many signals to decompose our data to.  This is called model order selection.  If we choose a number that is too... Read more