Use of Deep Learning Networks in Recognition of Visual Objects

Graduate papers

Search graduate papers

Description of the graduate paper

Form of studies	Bachelor
Title of the study programm	Computer Systems
Title in original language	Dziļās apmācības tīklu izmantošana vizuālu tēlu pazīšanai
Title in English	Use of Deep Learning Networks in Recognition of Visual Objects
Department	12300 Institute of Applied Computer Systems
Scientific advisor	Agris Ņikitenko
Reviewer	LDI pētnieks J.Bicāns
Abstract	Darbā tika pētīta mašīnmācīšanās apakšnozare, dziļie neironu tīkli. Lai gūtu nepieciešamās pamatzināšanas par neironu tīklu uzbūvi un darbības principiem, sākotnēji tika apskatīti vienkāršākie neironu tīkli – perceptroni. Pēc tam tika pētīti dziļie neironu tīkli un citi mašīnmācīšanās algoritmi, to ietekmējošie parametri un kādā veidā tos ir iespējams apmācīt. Rezultātā autors ieguva zināšanas neironu tīklu nozarē, kā arī apgūva vairākus mašīnmācīšanās algoritmus, kurus var izmantot tālākajā izpētē. Ir iegūts darbojošs modelis, kas spēj atpazīt un klasificēt BSL zīmju valodu, kā arī tika izstrādāta mobilā lietotne, kas nodrošina lietotāja saskarsmi ar izstrādāto sistēmu.
Keywords	DAUDZSLĀŅU PERCEPTRONS, UZRAUDZĪTA APMĀCĪBA, DZIĻIE NEIRONU TĪKLI, SVM, KNN
Keywords in English	MULTI-LAYER PERCEPTRON, SUPERVISED LEARNING, DEEP NEURAL NETWORKS, SVM, KNN
Language	lv
Year	2019
Date and time of uploading	27.05.2019 15:07:46