Kohonen Self Organizing Map (SOM)
This project implements Kohonen Self Organizing Map (SOM) and trains the network whilst investigating the effects of initial hyperparameter settings.
The Kohonen SOM provides a data visualization technique which helps to understand high dimensional data by reducing the dimensions of data to a map. SOM also represents clustering concept by grouping similar data together.
Unlike other learning technique in neural networks, training a SOM requires no target vector. A SOM learns to classify the training data without any external supervision.
Within this notebook, several concepts are discussed, from algorithm complexity and speed, to hyperparameter sensitivity analysis as shown below: