aalto1 untyped-item.component.html
Magnonic convolutional neural network
Loading...
Files
Aalto login required (access for Aalto Staff only).
URL
Journal Title
Journal ISSN
Volume Title
Perustieteiden korkeakoulu |
Bachelor's thesis
Electronic archive copy is available locally at the Harald Herlin Learning Centre. The staff of Aalto University has access to the electronic bachelor's theses by logging into Aaltodoc with their personal Aalto user ID. Read more about the availability of the bachelor's theses.
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.
Authors
Date
Department
Major/Subject
Mcode
SCI3103
Degree programme
Language
en
Pages
24
Series
Abstract
The computational requirements of operating machine learning algorithms are increasing faster than the performance improvement provided by new fabrication processes. This has lead to an increase of data centres size and, subsequently, of energy consumption and operational costs. To allow machine learning algorithms to develop further, new technologies that have been designed to increase computational efficiency are necessary.
This thesis proposes to harness magnonics to develop a convolutional encoder capable of performing computations more efficiently. Such a device could then be scaled to achieve effective parallel matrix-vector multiplication computations; thus easing machine learning algorithms operations.
Specifically, the magnonic convolutional neural network was designed, fabricated and tested using structured YIG thin film. The accuracy achieved in image recognition tasks are higher than 90%, which is comparable to software-based approach. These results show high potential and feasibility of magnonics as a novel technology for efficient neuromorphic processing.