Handwritten Digit Recognition Using Perceptron Neural Network

Introduction What is an artificial neural network and how does it work? Artificial neural network have been developed from generalizations of neural biology model, based on the assumptions that (a) information processing occurs at many simple elements called neurons, (b) signals are passed between neurons over connection links, (c) each connection link has an associated weight, which, in a typical neural net, multiplies the signal transmitted, and (d) each neuron applies an activation function to its net input to determine its output signal. A neural network is characterized by its pattern of connections between the neurons, its method of determining the weights on the connections, and its activation function. A neural net consists a large number of neurons, connected to other neurons by means of directed communication links, each with an associated weight. The weights represent information being used by the net to solve a problem. Each neuron has an internal state, called its activation or activity level, which is a function of the inputs it has received. When the resulting function from the given inputs exceed a predefined threshold, the neuron “fires” (send signal to other neurons). The neural net just been described can be modeled in the following way: Given a neuron Y that receives inputs from neurons X1, X2, and X3. The weights on the connections from X1, X2, and X3 to neuron Y are w1, w2, and w3. The input to neuron Y is the sum of the weighted signals from neurons X1, X2, and X3 given in the following equation: y_in = w1x1 + w2x2 + w3x3 The activation function of y is then computed based from the input value for neuron Y. Figure 1 illustrates the model of our simple neural net (typically called single-layer neural network).