prikaz prve stranice dokumenta Primjena rekurentnih neuronskih mreža u analizi programskog koda
  Download
PDF 16.76 MB
master's thesis
Primjena rekurentnih neuronskih mreža u analizi programskog koda
2021. urn:nbn:hr:126:855929
Jovanović, Antonio
Josip Juraj Strossmayer University of Osijek
Department of Mathematics
Chair of Applied Mathematics and Computer Science
Computer Science and Machine Learning Research Group

Cite this document

Jovanović, A. (2021). Primjena rekurentnih neuronskih mreža u analizi programskog koda (Master's thesis). Osijek: Josip Juraj Strossmayer University of Osijek, Department of Mathematics. Retrieved from https://urn.nsk.hr/urn:nbn:hr:126:855929

Jovanović, Antonio. "Primjena rekurentnih neuronskih mreža u analizi programskog koda." Master's thesis, Josip Juraj Strossmayer University of Osijek, Department of Mathematics, 2021. https://urn.nsk.hr/urn:nbn:hr:126:855929

Jovanović, Antonio. "Primjena rekurentnih neuronskih mreža u analizi programskog koda." Master's thesis, Josip Juraj Strossmayer University of Osijek, Department of Mathematics, 2021. https://urn.nsk.hr/urn:nbn:hr:126:855929

Jovanović, A. (2021). 'Primjena rekurentnih neuronskih mreža u analizi programskog koda', Master's thesis, Josip Juraj Strossmayer University of Osijek, Department of Mathematics, accessed 01 June 2023, https://urn.nsk.hr/urn:nbn:hr:126:855929

Jovanović A. Primjena rekurentnih neuronskih mreža u analizi programskog koda [Master's thesis]. Osijek: Josip Juraj Strossmayer University of Osijek, Department of Mathematics; 2021 [cited 2023 June 01] Available at: https://urn.nsk.hr/urn:nbn:hr:126:855929

A. Jovanović, "Primjena rekurentnih neuronskih mreža u analizi programskog koda", Master's thesis, Josip Juraj Strossmayer University of Osijek, Department of Mathematics, Osijek, 2021. Available at: https://urn.nsk.hr/urn:nbn:hr:126:855929

Please login to the repository to save this object to your list.
Metadata
TitlePrimjena rekurentnih neuronskih mreža u analizi programskog koda
Title (english)Application of Recurrent Neural Networs in Code Analysis
AuthorAntonio Jovanović
MentorDomagoj Matijević (mentor)
Committee memberKristian Sabo (predsjednik povjerenstva)
Committee memberDomagoj Matijević (član povjerenstva)
Committee memberDomagoj Ševerdija (član povjerenstva)
GranterJosip Juraj Strossmayer University of Osijek
Department of Mathematics
(Chair of Applied Mathematics and Computer Science)
(Computer Science and Machine Learning Research Group)
Osijek
Defense date and country2021-09-23, Croatia
Scientific / art field,
discipline and subdiscipline		NATURAL SCIENCES
Mathematics
Mathematical Logic and Accounting
Abstract
U ovom radu definiramo unaprijeđne i rekurentne neuronske mreže koje se koriste za rješavanje zadataka strojnog učenja. Neuronske mreže su sofisticirani model računanja za koji se može izraditi efikasan algoritam treniranja, procesa kojim neuronska mreža uči kako reprezentirati podatke uzorkovane iz neke nepoznate distribucije. U prva dva poglavlja rada definiramo razne oblike neuronskih mreža i dajemo primjere problema koje oni rješavaju. Naglasak se stavlja na LSTM arhitekturu, enkoder-dekoder arhitekturu i razne mehanizme pažnje. U trećem poglavlju motiviramo automatizaciju analize programskog koda. Dajemo pregled radova koji pokazuju kako primijeniti rekurentnu neuronsku mrežu za određivanje kratkog opisa funkcionalnosti isječaka programskog koda te klasifikaciju programskog koda.
Abstract (english)
In this paper we define feedforward and recurrent neural networks that are used to solve machine learning problems. Neural networks are a sophisticated computational model with an efficient training algorithm, a process where the networks learns how to describe data sampled from an unknown probability distribution. In the first two chapters, we define different types of neural networks and give examples of problems they can solve. We put emphasis on the LSTM architecture, encoder-decoder architectures and various attention mechanisms. In the third chapter, we motivate automation of program analysis. We review papers that show how to apply recurrent networks to summarize and caption code snippets and to classify programs.
Keywords
Keywords (english)
Languagecroatian
URN:NBNurn:nbn:hr:126:855929
Study programmeTitle: Mathematics; specializations in: Financial and Statistical Mathematics, Mathematics and Computer Science, Industrial and Applied Mathematics
Course: Financial and Statistical Mathematics
Study programme type: university
Study level: graduate
Academic / professional title: magistar/magistra matematike (magistar/magistra matematike)
Type of resourceText
File originBorn digital
Access conditionsOpen access
Terms of useLicence In copyright icon
Created on2021-09-24 08:14:22