Primjena umjetnih neuronskih mreža za predviđanje brzine vjetra

undergraduate thesis

• Title (english): Application of artificial neural networks for wind speed prediction
• Author: Fran Medved
• Mentor: Krešimir Trontl (mentor)
• Committee member: Mario Matijević (član povjerenstva)
• Committee member: Siniša Šadek (član povjerenstva)
• Committee member: Krešimir Trontl (predsjednik povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-09-13, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

U ovom radu radi se pregled i istražuje se rad s umjetnim neuronskim mrežama, od toga kako je napravljen umjetni neuron do kompleksnih neuronskih mreža. Napravljen je pregled različitih vrsta neurona, objašnjena je uloga prijenosne funkcije te su navedene prednosti i mane različitih arhitektura neuronskih mreža
Nadalje je pokazano kako neuronska mreža uči te je objašnjen algoritam propagacije pogreške unazad.
Drugi dio rada se bavi programskom implementacijom modela, koristeći LSTM arhitekturu. Napravljena je podjela ulaznih podataka na setove za treniranje, provjeru i testiranje te je implementirano učenje modela na temelju tih podataka. Na kraju su uspoređeni razni izlazi modela.

Abstract (english)

This paper provides an overview and explores the working principles of artificial neural networks, ranging from the creation of artificial neurons to complex neural network architectures. Various types of neurons are reviewed, emphasizing the role of activation functions, and the advantages and limitations of different neural network architectures are discussed.
Furthermore, the learning process of neural networks is demonstrated, with a detailed explanation of the backpropagation algorithm.
The second part of the paper focuses on the implementation of a model using the LSTM architecture. The input data is divided into training, validation, and testing sets, and the model is trained based on these datasets. Finally, a comparison of different model outputs is presented.

• Keywords: neuron; neuronska mreža; prijenosna funkcija; gradijent; gradijentni spust; funkcija pogreške; stopa učenja; propagacija pogreške unazad; LSTM neuronska mreža; pretreniranost; generalizacija
• Keywords (english): neuron; neural network; activation function; gradient; gradient descent; error function; learning rate; backpropagation; LSTM neural network; overfitting; generalization
• Language: croatian
• URN:NBN: urn:nbn:hr:168:452856
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-10-11 09:47:40