CS401
Machine Learning and Neural Networks

A very gentle introduction.

• Download our Artificial Neural Network Simulator, for Microsoft Excel  (at your won risk) by Adrian Trenaman and Diarmuid O'Donoghue.

This simulator is colour coded, where each cluster of rows displays the state of the network at a different time. See how the ANN learns from its mistakes. The tabs allow you to jump to the Adaline and (Simple) Perceptron, which are different models of neurons. This simulator also shows why this Perceptron is called Simple!

For a completely different type of Neural Network download our Hopfield Neural Network Simulator, for Microsoft Excel. From the diagram you can see that it tries to keep "alive" the maximum number of neurons, depending on the connections between them.
   
 

Topics:

• Machine Learning, Training, Target Function, Functional Approximation, intro-notes.
• Decision Trees, ID3, C4.5, C5, Overfitting, Entropy, id3-notes.
• Instance based (lazy) learning, K Nearest Neighbours, knn-notes.
• Neurons, Feed-Forward Neural networks, Adaline, Perceptron, Multi-Layer Perceptron, Linear Separability
• SVM (Support Vector Machines - see).

Text:
"Machine Learning" by Tom Mitchell, McGraw-Hill, 1997 -  including his lecture slides.
"Artificial Neural Networks: Theory and Applications" by Dan Patterson.

 

Diarmuid O'Donoghue. Sept. 2004.