Identifikacija tipova 1D-signala dubokog učenja

master's thesis

• Title (english): Classification of 1D-Signal Types Using Deep Learning
• Author: Filip Floreani
• Mentor: Mile Šikić (mentor)
• Committee member: Mile Šikić (predsjednik povjerenstva)
• Committee member: Damir Seršić (član povjerenstva)
• Committee member: Krešimir Križanović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-07-04, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

De-novo sastavljanje genoma proces je temeljen na preklapanju i analizi kratkih genetskih očitanja. Uslijed raznolikih tehničkih izazova, određene vrste lažnih očitanja i preklapanja mogu također biti sintetizirane, što uvelike otežava is- pravno sastavljanje originalne sekvence. Jedna od metoda za otkrivanje takvih preklapanja je i pretvorba istih u 1D-signal, pomoću kojih se onda može jed- nostavnije zaključiti koji tip preklapanja predstavlja. U okviru ovog rada, pred- stavljeno je nekoliko metoda duboko učenja za klasifikaciju ovih signala, točnije 1D-konvolucijske mreže, povratne mreže, kao i primjena autoenkodera. Dodatno, prikazana je i njihova primjena na stvarnim podacima, kao i usporedba usp- ješnosti.

Abstract (english)

The de novo genome assembly process is based on overlapping and analyzing short reads of genetic information. Due to various technical challenges, certain types of false overlaps can also be generated, which greatly impedes successful reconstruction. One of the methods for detecting such overlaps is by generating a 1D-signal for each read, which can then be used to determine its exact overlap type. This thesis proposes several deep learning methods for classifying these signals, including 1D-convolutional & recurrent networks, as well as autoencoders. A detailed comparison of their application on real-world data is also included.

• Keywords: bioinformatika; sastavljanje genoma; lažna preklapanja; duboko učenje
• Keywords (english): bioinformatics; sequence assembly; false overlaps; deep learning
• Language: english
• URN:NBN: urn:nbn:hr:168:825125
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2020-01-07 17:56:17