I seem to have learnt a lot of theory relating to machine learning algorithms. This seems like the right time to experiment with my understanding of each of the algorithms. I would be implementing as many as possible from the Machine Learning by Tom Mitchell book.
The current implementation of decision trees is based on boolean attributes and classes. You can construct the decision tree using the best attribute selection as in the ID3 algorithm or by randomly choosing attributes. The random choice of attributes may give you bigger trees (which are not always good).
This implementation of neural networks is a close representation of the MLPClassifier of Scikit-Learn. The main algorithm implemented is the Back-Propagation algorithm. I am also working on a parallelization of the code over the cores in the system for faster implementation and weight tuning.
Check out the code or contact me and I’ll help you out.