Artificial Neural Network is the excited and hottest topic in this generation. The Artificial Neural Network is depending on the Study of the Neural Network. It provides us a general, practical knowledge for learning real-valued vector valued target functions. This ANN is inspired by the study of interconnected Neurons present in the human body. These Neurons will pass the message to one part of the body to another part of the body with in the seconds.
The Human brain consist of ~10 11 Neurons. and ~ 104-5 are connection per neurons.
One type of ANN system is based on a unit called Perceptron. The perceptron is a single Layer Neural Network.
Perceptron takes the input as real valued vectors and gives the output as 1 if the result is greater than some Threshold value else it will give the output as -1, like let’s take example
Input – x
Output – O (x1, . . . , xn)
Where, each wi is a real-valued constant, or weight, that determines the contribution of input xi to the perceptron output.
w0 is a threshold that the weighted combination of inputs w1x1 + . . . + wnxn must surpass in order for the perceptron to output a 1.
Sometimes, the perceptron function is written as,
Learning a perceptron involves choosing values for the weights w0 , . . . , wn . Therefore, the space H of candidate hypotheses considered in perceptron learning is the set of all possible real-valued weight vectors.
Perceptron can represent all primitive Boolean functions like OR, AND, NOR, NAND.
Perceptron will not represent the Boolean function like XOR.
Drawbacks of Perceptron:
The perceptron will only work when training vectors are linearly separable. It will fail when the training vectors are not linear.