Algoritam rojeva čestica u učenju neuronske mreže

undergraduate thesis

• Title (english): Particle swarm algorithm in neural network learning
• Author: Ivan Čović
• Mentor: Dubravko Majetić (mentor)
• Committee member: Bojan Jerbić (predsjednik povjerenstva)
• Committee member: Dubravko Majetić (član povjerenstva)
• Committee member: Josip Kasać (član povjerenstva)
• Committee member: Petar Ćurković (član povjerenstva)
• Granter: University of Zagreb; Faculty of Mechanical Engineering and Naval Architecture; Zagreb
• Defense date and country: 2019-09-26, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing; Artificial Intelligence

Abstract

U ovom radu opisani su algoritmi učenja neuronske mreže povratnim prostiranjem pogreške i algoritam optimizacije rojem čestica (PSO). Obrađena su dva klasična primjera (XOR i Iris set) u učenju neuronskih mreža, pomoću dvije navedene metode učenja neuronskih mreža. Uspoređuju se svojstva algoritama nakon izvođenja primjera kao što su: konvergencija greške, veličina greške, te brzina izvođenja algoritma u zadanim parametrima. Uz to, u uvodu se opisuje učenje i glavne značajke neuronskih mreža, a zatim način učenja i rada zadane mreže. Na kraju je izveden zaključak, na temelju obrađenih primjera.

Abstract (english)

This paper describes the neural network training algorithms by error backpropagation and the particle swarm optimization (PSO). Two classical examples (XOR and Iris set) in neural network training were analyzed using the two methods. The properties of algorithms after executing examples are compared, such as: error convergence, magnitude of error, and the velocity of algorithm execution in given parameters. In addition, the introduction describes the training and main features of neural networks, followed by how the default network is trained and operated. Finally, a conclusion is drawn, based on the examples discussed.

• Keywords: neuronske mreže; umjetna inteligencija; umjetni neuron; učenje neuronske mreže; težinski faktori; povratno prostiranje pogreške; optimizacija rojem čestica
• Keywords (english): neural networks; artificial intelligence; artificial neuron; training neural network; neural network weights; error backpropagation; particle swarm optimization
• Language: croatian
• URN:NBN: urn:nbn:hr:235:655191
• Study programme: Title: Mechanical Engineering; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka strojarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka strojarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2019-09-27 08:03:32