Κατηγοριοποίηση πλοίων με την χρήση νευρωνικών δικτύων
Ship classification with neural networks
Ευθυμίου, Άγγελος Χρ.
KeywordsΑλγόριθμοι ; Νευρωνικά δίκτυα ; Πλοία ; Αρχιτεκτονική ηλεκτρονικών υπολογιστών ; Neural networks (Computer science) ; Fuzzy
The purpose of this paper is to create and use a neural network, which will be used to classify different types of ships, based on the characterized movement. Nowadays, many architectures of prediction models are used and we could have the opportunity to present them. Fuzzy means, which is a powerful algorithm and based on radial basis action of neural network architecture, will be the core component for the neural network. In order to achieve the correct results, we will use one vs all algorithm, which is used for classification problems, when we have more than two classes as outputs. All data have received from the University of Piraeus antenna, which can encapsulate data around the area of Argosaronikos sea. Firstly, we will train the neural network with all the available datasets, without any process of data. As inputs, we will give the transportation, the speed and the acceleration of every class. After that, we will try to improve the network by processing the input data, in order to get rid of the noise and then we will retrain the network. If the result is successful, we will check the robustness of the network via changing some of the parameters, in order to observe, if the response of the network, will remain in good rate.