Ispitivanje učinkovitosti automatizacije grafičke pripreme primjenom strojnog učenja

master's thesis

• Author: Alen Okanović
• Mentor: Dubravko Banić (mentor)
• Committee member: Dubravko Banić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Graphic Arts; Zagreb
• Defense date and country: 2013-10-31, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Graphic Technology

Abstract

Strojno učenje je znanstvena disciplina koja se bavi razvojem algoritama koji računalu omogućuju da na temelju empirijskih podataka iz senzora ili baze podataka razvije autonomno ponašanje. Rad istražuje mogućnosti automatizacije grafičke pripreme primjenom tehnika strojnog učenja kojima bi se ljudski rad i donošenje odluka prepustilo umjetnoj inteligenciji. Rad adresira problem kategorizacije velikih količina dokumenata čija je obrada ljudkim radom zbog velikog opsega vrlo nepraktična, vremenski zahtjevna i podležna greškama. Radom se ispituju i uspoređuju različiti pristupi kojima bi se smanjila potreba za ljudskim radom, a kategoriziranje prepustilo računalu s namjerom ubrzanja proizvodnje i smanjenjem troškova pripreme. Uvodni dio sadrži informativni pregled strojnog učenja, zanimljive dosege istraživanja iz tog znanstvenog područja i razmatraju se razni pristupi rješavanju problema istraživanja. U središnjem dijelu ispituje se funkcionalnost, pouzdanost i opravdanost automatizacije strojnim učenjem te će se ovisno o rezultatima ispitivanja detaljno razložiti prednosti i nedostaci rješenja. Pretpostavlja se da je priprema dokumenata ovakvom vrstom automatizacije najmanje jednako kvalitetna, ali višesturko brža od pripreme korištenjem ljudskog rada.

Abstract (english)

Machine learning deals with the development of algorithms that allows a computer to develop autonomous behavior based on empirical data from sensors or databases. This paper explores the possibilities of the prepress automation with application of machine learning techniques in which decision-making would be left to artificial intelligence. Paper addresses the problem of categorizing large amounts of documents whose processing by workers is highly impractical due to its large scale. Besides that, it is time-consuming and it could be subject of human errors. This paper examines and compares different approaches that reduce the need for human labor and categorizing is left to computer program with the intention of speeding up production and reducing the cost of prepress. The introductory section contains an informative overview of machine learning, interesting achievements in this scientific field and it discusses various approaches to machine learning automation. Main section examines reliability and reasons for machine learning automation and, dependent on the research results, it lays down the advantages and disadvantages of such solutions. It is assumed that the preparation of documents with this kind of automation is at least as good, but many times faster than preparations using human labor.

• Keywords: grafička priprema; automatizacija; kategorizacija dokumenata
• Keywords (english): prepress; automatization; document classification
• Language: croatian
• URN:NBN: urn:nbn:hr:216:921292
• Study programme: Title: Graphic technology; specializations in: Graphic technology, Design of graphic products; Course: Graphic technology; Study programme type: university; Study level: graduate; Academic / professional title: sveučilišni/a magistar/magistra inženjer/inženjerka grafičke tehnologije (sveučilišni/a magistar/magistra inženjer/inženjerka grafičke tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2024-04-17 09:14:07