| | SLO | ENG | Cookies and privacy

Bigger font | Smaller font
First page > Show document

Show document Help

Title:Podatkovni kanjoni, pristop strojnega učenja za potrebe razložljive umetne inteligence : doktorska disertacija
Authors:ID Žlahtič, Bojan (Author)
ID Kokol, Peter (Mentor) More about this mentor... New window
Files:.pdf DOK_Zlahtic_Bojan_2023.pdf (3,26 MB)
MD5: 062B25EDB220C1A4F8203C1CA7AA64E8
 
Language:Slovenian
Work type:Doctoral dissertation
Typology:2.08 - Doctoral Dissertation
Organization:FERI - Faculty of Electrical Engineering and Computer Science
Abstract:Z uporabo algoritmov strojnega učenja je mogoče izvesti zapletene analize in pridobiti globlje vpoglede na osnovi obsežnih količin podatkov, kar presega človeške zmožnosti. Navedena značilnost je ključni dejavnik, zaradi katerega je strojno učenje vpeljano v številne domene. Kljub številnim prednostim ni vedno možno integrirati strojnega učenja na določena področja, predvsem zaradi tega, ker se za naprednimi metodami pogosto skrivajo modeli tipa črne skrinje. Ti modeli uporabnikom ne omogočajo vpogleda v logiko njihovega odločanja, kar lahko predstavlja oviro v kontekstih, kjer so odločitve kritične in lahko napačna odločitev vodi v resne posledice. Z namenom ublažiti te problematike smo razvili metodo strojnega učenja, temelječo na naravnem pojavu rečnih kanjonov. Ta pojav lahko vizualiziramo v digitalni grafični obliki, kar omogoča intuitiven prikaz logike odločanja. Rezultat je model strojnega učenja, ki generira globinske slike gibanja podatkov za posamezen razred. V teh slikah je pripadnost posamezne instance kanjonu prikazana s pomočjo barvno kodiranih grafov. Podatkovni kanjoni se zaradi svojih lastnosti in metodologije lahko uporabljajo za potrebe razložljive umetne inteligence, bodisi samostojno ali kot dopolnilni mehanizem drugim pristopom strojnega učenja.
Keywords:razložljiva umetna inteligenca, strojno učenje, klasifikacija, razložljivost, zaupanje
Place of publishing:Maribor
Place of performance:Maribor
Publisher:[B. Žlahtič]
Year of publishing:2023
Number of pages:VIII, 103 str.
PID:20.500.12556/DKUM-85798 New window
UDC:004.85.021:004.6(043.3)
COBISS.SI-ID:174856451 New window
Publication date in DKUM:05.12.2023
Views:467
Downloads:91
Metadata:XML DC-XML DC-RDF
Categories:KTFMB - FERI
:
ŽLAHTIČ, Bojan, 2023, Podatkovni kanjoni, pristop strojnega učenja za potrebe razložljive umetne inteligence : doktorska disertacija [online]. Doctoral dissertation. Maribor : B. Žlahtič. [Accessed 12 April 2025]. Retrieved from: https://dk.um.si/IzpisGradiva.php?lang=eng&id=85798
Copy citation
  
Average score:
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
(0 votes)
Your score:Voting is allowed only for logged in users.
Share:Bookmark and Share


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

Licences

License:CC BY-NC-ND 4.0, Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International
Link:http://creativecommons.org/licenses/by-nc-nd/4.0/
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:12.09.2023

Secondary language

Language:English
Title:Data canyons, a machine learning approach for interpretable artificial intelligence
Abstract:With machine learning algorithms avast amounts of data can be analyzed to gain profound insights beyond human capabilities. This quality is a key reason for incorporating machine learning across numerous domains. Although machine learning has various benefits aincorporating it into certain areas can be difficult due to the blackbox nature of advanced methods. These models do not grant users insight into their decision-making logic aposing challenges in contexts where decisions are critical and an erroneous decision might result in severe consequences. To address these issues awe developed a machine-learning method based on the natural phenomenon of river canyons. This phenomenon can be visualized in a digital graphical format aoffering an intuitive representation of decision-making logic. The outcome is a machine learning model that produces depth maps of data movement for individual classes. Within these maps athe affiliation of a particular instance to a canyon is denoted by color-coded graphs.
Keywords:explainable artificial intelligence, machine learning, classification, explanation, trust


Comments

Leave comment

You must log in to leave a comment.

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

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