| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font

Show document

Title:Klasifikacija časovnih vrst s konvolucijskimi nevronskimi mrežami : magistrsko delo
Authors:Kavran, Domen (Author)
Lukač, Niko (Mentor) More about this mentor... New window
Files:.pdf MAG_Kavran_Domen_2020.pdf (10,95 MB)
MD5: 69887BA0950FDECFD0B024F1071691DC
Work type:Master's thesis/paper (mb22)
Typology:2.09 - Master's Thesis
Organization:FERI - Faculty of Electrical Engineering and Computer Science
Abstract:V magistrskem delu predstavimo klasifikacijo časovnih vrst z uporabo konvolucijskih nevronskih mrež. Klasifikacija je izvedena nad časovno-frekvenčnimi predstavitvami časovnih vrst, ki so pridobljene z različnimi metodami časovno-frekvenčne analize. Zasnovali smo več arhitektur konvolucijskih nevronskih mrež za klasifikacijo časovnih vrst. Optimizacijski algoritmi za učenje konvolucijskih nevronskih mrež so uporabljali napredno izgubno funkcijo, imenovano žariščna izguba. Za najuspešnejšo metodo izračuna časovno-frekvenčnih predstavitev časovnih vrst se je izkazala zvezna valčna transformacija, s katero smo dosegli povprečno natančnost klasifikacije 90,07 %. Združitev različnih časovno-frekvenčnih predstavitev je izboljšala povprečno natančnost klasifikacije na 92,01 %.
Keywords:klasifikacija, globoko učenje, konvolucijske nevronske mreže, časovne vrste, časovno-frekvenčna analiza
Year of publishing:2020
Place of performance:Maribor
Publisher:[D. Kavran]
Number of pages:X, 64 f.
COBISS_ID:27221763 New window
Categories:KTFMB - FERI
Average score:(0 votes)
Your score:Voting is allowed only for logged in users.
AddThis uses cookies that require your consent. Edit consent...

Hover the mouse pointer over a document title to show the abstract or click on the title to get all document metadata.


License:CC BY-NC-ND 4.0, Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International
Description:The most restrictive Creative Commons license. This only allows people to download and share the work for no commercial gain and for no other purposes.
Licensing start date:05.05.2020

Secondary language

Title:Time series classification based on convolutional neural networks
Abstract:This master’s thesis presents time series classification using convolutional neural networks. Classification is performed on time-frequency representations of time series, which are obtained by using different time-frequency analysis methods. Several convolutional neural network architectures for time series classification were designed. Optimization algorithms for learning convolutional neural networks used advanced loss function, called focal loss. The most successful method for computing time-frequency representations of time series has proven to be a continuous wavelet transform, which achieved an average classification accuracy of 90,07 %. Combining various time-frequency representations increased average classification accuracy to 92,01 %.
Keywords:classification, deep learning, convolutional neural networks, time series, time-frequency analysis


Leave comment

You have to log in to leave a comment.

Comments (0)
0 - 0 / 0
There are no comments!

Logos of partners University of Maribor University of Ljubljana University of Primorska University of Nova Gorica