Please use this identifier to cite or link to this item:
Author: Χριστοφορίδης, Αριστείδης
Title: A novel evolutionary algorithm for hierarchical neural architecture search
Date Issued: 2021
Department: Πρόγραμμα Μεταπτυχιακών Σπουδών στην Τεχνητή Νοημοσύνη και Αναλυτική Δεδομένων
Supervisor: Μαργαρίτης, Κωνσταντίνος
Abstract: In this thesis, we propose a new neural architecture search algorithm that performs network discovery in global search spaces. We introduce a novel network representation that organizes the topology on multiple hierarchical levels of varying abstraction and develop an evolution based search process that exploits this structure to explore the search space. Our approach involved a curation system that selects well performing network components and uses them in subsequent generations to build better networks. Next, we investigate how the proposed method performs on different types of data. First, we apply our method on an activity recognition time series dataset and manage to discover a topology with impressive performance. We also test the method on two image classification datasets, Fashion-MNIST and NAS-Bench-101 and achieve accuracies of 93.2% and 94.8% respectively in a small amount of time.
Keywords: Deep neural networks
Neural architecture search
Information: Διπλωματική εργασία--Πανεπιστήμιο Μακεδονίας, Θεσσαλονίκη, 2021.
Rights: CC0 1.0 Παγκόσμια
Appears in Collections:ΠΜΣ στην Τεχνητή Νοημοσύνη και Αναλυτική Δεδομένων (Μ)

Files in This Item:
File Description SizeFormat 
ChristoforidisAristeidisMsc2021.pdf1.56 MBAdobe PDFView/Open
ChristoforidisAristeidisMsc2021present.pdfΠαρουσίαση1.58 MBAdobe PDFView/Open

This item is licensed under a Creative Commons License Creative Commons