I have been working on writing an analog artificial neural network with hidden layers and back propagation of error. I have written a single perceptron model which works alright with an R^2 of 0.64 consider the nature of the input data. I used the concepts developed in this first program and applied it the multi layered perceptron model. However, I am having difficultly with the accuracy of prediction. I have been playing with a few free variables and it has seem to increase the accuracy but my progression is slow.
I was wondering if you have attempted anything along artificial neural networks.
C'était ton utilisation de trémas dans ta langue inventée, je pense. Mais si c’est basé sur le Français, ça explique pourquoi je n’ai pas eu grande difficulté à la comprendre.
J’habite sur la rive sud, mais je suis aussi souvent en Montréal grâce à l’école.