This repository is dedicated to machine learning study, allways focused on implementing code, focused on the needs of Computer Engennering graduation courses, but in the hope it can be usefull to anyone willing to study machine learning.
My goal is to develop Java code that implements machine learning algorithms targeting education. The code must be clean, self explanatory, walking step by implemented algorithms, in order to be used as material for machine learning study. So usually the code you can find here is not made to be optimized, rather, it is made to be easy to understand.
There are several Java machine learning libraries available on the internet. As they target easy of use and high performance, their code brings extra complexity. We will avoid using them, for the reasons stated in the above paragraph. Of course some exceptions should be made for good and simple support libraries, like i.e. Apache math library with linear algebra.
The CIFAR-10 dataset (https://www.cs.toronto.edu/~kriz/cifar.html) will be used for trainning and testing of image classification and recognization, in several approaches, like, regression, svm, neural networks, convolutional networks, and so on.
At the present time 2 algorithms that can be used in image classification are available: the nearest neighbourg (nn) and the k-nearest neighbourgh (knn).
I recomend start by reading the wiki, but, please, don't forget this is a work in progress, and at the very beggining.