Inteligentní metody na bázi strojového učení pro klasifikaci a identifikaci objektů zájmu: laboratorní úloha

Abstract

The aim of the master thesis is to design exemplary laboratory tasks to introduce students to data classification with neural networks. The individual tasks deal with data classification. The first assignment is about the basic method of classification using perceptron. Other tasks focus on the neural network optimization method using genetic algorithms and the next assignment deals with the use of convolutional neural networks for classification of one-dimensional acoustic signals and two-dimensional images. For each task, data is created or loaded to train the neural networks, then create the network and train it under different settings. In the last step of the lab tasks, the algorithms are validated, and the results are analysed. All the algorithms of the subsections were programmed in MATLAB using Live Script technology.