Anselme Russel Affane Moundounga I would recommend you look at the source code of numerical algorithms in the GNU scientific library, those contain state-of- the-art implementations in C/C++ for algorithms found in NR3, but optimized + adapted for edge/real world cases. It's a great exercise to compare your own code against it as well.
In the GSL routines, I've not find routines about the hidden Markov Models, Expectation-maximization, Gaussian Mixtures Models, K-means clustering and many others used in machine learning and classification methods.