[home]

# Machine Learning Applets

``...w0ah!''

Alex S.

[Least Squares]: A very simple implementation of Least Squares method; both primal and dual ridge regression (plot points with a mouse, and let the applet fit a line);
Wed May 21 02:00:34 EDT 2008

[Least Squares Discriminator]: Left click adds `x', Right click adds `o'. Applet finds the best fitting separating line. Type any key to reset.
Tue Jun 3 01:22:13 EDT 2008

[Least Squares for Polynomials]: A very simple implementation of Least Squares method; both primal and dual ridge regression to find best fitting polynomial.
Tue Jun 23 07:32:08 EDT 2009

[NonLinear Least Squares]: The applet will use ridge regression to fit exponential, e.g, y=b*e^(a*x), and a power, e.g. y=b*x^a, curves.
Mon Aug 31 23:35:08 EDT 2009

[Perceptron_Discriminator]: Plot 'x' and 'o's. When you click, applet starts with random weights and uses gradient descent (learning rate 0.1) to train a perceptron [sigmoid, 1/(1+exp(-a)) function] to classify the points. The applet does 10000 iterations, and doesn't always succeed.
Thu Jun 26 02:21:48 EDT 2008

[Nearest_Neighbor]: A very simple implementation of kNN algorithm, with k=1. The gist is that you plot 'x', and 'o', and the applet figures out how things would be classified based on whether x or o is closest.
Mon Jun 16 01:14:57 EDT 2008

[K-Means]: Click to plot a bunch of points. The applet will assign color and character to each point, based on k-Means algorithm, with k=4.
Mon Jun 16 03:00:59 EDT 2008

[Hopfield Network]: Implements a pattern recognizer using Hebbian learning, and Hopfield recurrent network.
Thu Jul 16 00:51:13 EDT 2009

[home]