Part 0 - Coding a neural network from scratch in C++11
Line by line code session of a version of the multi layer perceptron, the most popular type of the neural network model. The implementation use C++ 11 functional features and STL. No third part libraries are included, just using googletest framework as test engine. The project is build with cmake/make and it is shared on github.com for educational purposes. The final implementation is evaluated and its performance is compared with the results from another implementation, the Weka 3 (from Waikato University). Using the same data and the same parameters both implementations outputs the same results exaclty, regarding the differences of platform and approaches.
The backpropagation model is taken from the Simon Haikyn book, Neural Networks a Comprehensive Foundation.
code repository: https://github.com/doleron/neuralnet
part iii: • Video
part ii: • Video
part i: • Video
part 0: • Part 0 - Coding a neural network from scra...