02.04.06
Its finally over… for now!
Its finally over… what a sem… the most challenging, yet so far… but at least, now i understand what its mean by multilayer perceptron with a number of hidden nodes or neurons… to differentiate it from the basic of artificial neural network, the perceptron itself which consists of a linear combiner followed by a hard limiter… how to calculate the output of the neuron from the weighted input and make a comparison to the desired one based on the initial threshold level… and i can now calculate the error gradient and to do some weight correction… or to accelerate its learning process using sigmoidal activation function with hyperbolic tangent… or maybe i need to apply a constant or momentum term to its back propagation… argh!!! It’s still a far cry from Stealth or I, Robot..
to be continued…
