#ml

• Gaussian mixture

  A Gaussian mixture model fits an ellipse to each group of points, and each point belongs partly to every group. Expectation maximization fits it round by round, animated.

• k-means

  k-means splits points into k groups by repeating two steps, color each point by its nearest center and move each center to the middle of its points, animated round by round.

• Line fit

  Drag the points and the least-squares line refits as they move, with the squares it makes smallest drawn on the canvas.

• PCA

  Principal component analysis finds the direction the points spread out most. Drag the points and the directions, the ellipse and the share of the spread update as they move.

• SVM

  A linear support vector machine finds the widest gap between two colors of points. Drag the points and the gap refits, with rings on the few points that hold it in place.

All tags