Kohonen önszerveződő térképei az oktatásban

Abstract

A Kohonen network, also known as a Kohonen action map, is a type of artificial neural network that falls into the category of self-organizing networks. A Kohonen network was developed by Teuvo Kohonen in the 1980s and is one of the most widely used types of self-organizing networks. The goal of a Kohonen network is to group data into a twodimensional map based on their similarity. In a Kohonen network, data is provided as an input layer, and the output layer forms a two-dimensional grid in which each neuron belongs to a cell. The algorithm feeds data into the network through the input layer and assigns the data to the nearest neuron based on its characteristics. The output neurons then change their activity due to their influence on nearby neurons, causing the network to selforganize and group similar data into the same group. Kohonen networks are widely used for pattern classification, segmentation, and dimensionality reduction, as well as for data visualization. Kohonen networks use unsupervised learning, which allows the network to work efficiently on data without prior knowledge of the output results. This can also be used in education.

A Kohonen network, also known as a Kohonen action map, is a type of artificial neural network that falls into the category of self-organizing networks. A Kohonen network was developed by Teuvo Kohonen in the 1980s and is one of the most widely used types of self-organizing networks. The goal of a Kohonen network is to group data into a twodimensional map based on their similarity. In a Kohonen network, data is provided as an input layer, and the output layer forms a two-dimensional grid in which each neuron belongs to a cell. The algorithm feeds data into the network through the input layer and assigns the data to the nearest neuron based on its characteristics. The output neurons then change their activity due to their influence on nearby neurons, causing the network to selforganize and group similar data into the same group. Kohonen networks are widely used for pattern classification, segmentation, and dimensionality reduction, as well as for data visualization. Kohonen networks use unsupervised learning, which allows the network to work efficiently on data without prior knowledge of the output results. This can also be used in education.