MINING data

Train Support Vector Machine (SVM) with RBF kernel on part of IRIS dataset and classify the rest.
Uses two first values of a datapoint as x,y mapped with max/min boundary to width/height of canvas.

Using svm.js by github.com/karpathy.

Click for node and CTRL-click on two nodes to create an edge

 

 

Eigenvectors of the calc. Laplacian Matrix found with matrix.js by github.com/Stanley.

Learning XOR is the hello world of neural networks (NNs) using supervised learning. I've added a few more logic gates to the example code, including NOT that requires 1 instead of 2 neurons in the input and hidden layer. Also optimized/refactored it a little, and added own display of the truth table from the 2D-arrays with the training data (INPUT) and the ideal data (IDEAL). The Resilient Propagation (1992) is more efficient and modern than the Backpropagation (1970s) learning algorithm.


  

  

  

Based on example http://www.heatonresearch.com/fun/ann/xor using the Encog Javascript library.

Blue pill | Red pill Chaos | Blobs