For sometime now, I have been trying to code up all the algorithms that I understand in C++. This has helped me immensely in both sharpening my skills as a C++ programmer, as well as have given me a deeper understanding of what is going on. I am building it on OpenCV so there is a lot parallels between Matlab syntax and what I am doing and the fact it is relatively easy to make it portable across Windows and Linux. Some algos I am done with are Logistic Regression, Softmax, SVM (SMO), Naive Bayes and Backpropagation for Neural Nets. On the graphical side, I have Sumproduct and MaxSum. May be in a couple of months, I will also try to get running with deep learning algos, but to be frank, right now, I have no idea of Neural Nets besides the very basic Backpropagation.
P.S.: I understand there are libraries out there, mlpack, and others for ML in C++ with a number of great contributors and I am no comparisan.
P.S.S.: I have an MSEE.
[link][10 comments]