Neural network modeling on the example of the perceptron in the study of the classification of objects

Vasiliev Alexander Vladimirovich

The article is devoted to the neural network modeling using the perceptron as an example in the study of object classification. The objective of the study is to classify images of certain colors by distinguishing parameters and adjust the algorithm in such a way as to minimize erroneous classification. Like the human brain, neural networks are made up of a large number of connected elements that mimic neurons. Deep neural networks are based on such algorithms, thanks to which computers learn from their own experience, forming in the learning process multi-level, hierarchical representations of the world. The perceptron is the most primitive neural network that has two input and one output cell. Perceptrons make it possible to create a set of “associations” between input stimuli and the desired response at the output. In a biological context, this is the conversion, for example, of visual communication into physiological actions from motor neurons. Perceptrons are grouped as artificial neural networks: with one hidden layer; with threshold transfer function; with direct signal propagation. Perceptrons can be used in solving problems of binary classification, when the sample must be identified as belonging to one of the predefined two classes. In this research paper, the perceptron operation algorithm, the learning algorithm, and the algorithm correction were written.

Keywords:neural network, neurons, perceptrons, algorithm, function, model, weights.