Politehnika i dizajn, Vol. 7 No. 1, 2019.
Pregledni rad
https://doi.org/10.19279/TVZ.PD.2019-7-1-04
OVERVIEW OF ARTIFICIAL NEURAL NETWORK TECHNOLOGIES
Tin Kramberger
; Tehničko veleučilište u Zagrebu, Zagreb, Hrvatska
Bojan Nožica
; Tehničko veleučilište u Zagrebu, Zagreb, Hrvatska
Ivica Dodig
; Tehničko veleučilište u Zagrebu, Zagreb, Hrvatska
Davor Cafuta
; Tehničko veleučilište u Zagrebu, Zagreb, Hrvatska
Sažetak
Neural networks are being researched more and
more today. The reason for research lies in the
hardware that nowadays offers the ability to
process large amounts of data in real time. For
the successful operation and construction of the
neural network, one of great importance is the
activation function.
Activation function selection affects the speed and
quality of training the neural network itself. The
basic principles of the neural network after the
selection of activation functions are explained in
the paper. The basic principles of learning neural
networks are outlined, focusing on selecting the
optimization algorithm used to learn the neural
network.
Ključne riječi
Hrčak ID:
220655
URI
Datum izdavanja:
30.4.2019.
Posjeta: 1.798 *