Learn computer vision fundamentals with the famous MNIST data
Most challenging and funniest competition out of which I took part. Computer Vision projects' outputs are easily readable and understandable e.g each digit can be represented as a heatmap matrix or digits can be clustered and distinguished by using Principal Components Analysis which combines features into only 2 mixed features that can be easily plotted.
Support Vector Classifier occurs to be performing the best reaching 0.97521 but still, there is a lot to improve by learning C Neural Networks.