iGNNspector: a tool for graph neural network acceleration
Visualitza/Obre
Estadístiques de LA Referencia / Recolecta
Inclou dades d'ús des de 2022
Cita com:
hdl:2117/356323
Tipus de documentTreball Final de Grau
Data2021-06-30
Condicions d'accésAccés obert
Tots els drets reservats. Aquesta obra està protegida pels drets de propietat intel·lectual i
industrial corresponents. Sense perjudici de les exempcions legals existents, queda prohibida la seva
reproducció, distribució, comunicació pública o transformació sense l'autorització del titular dels drets
Abstract
This thesis is also part of a bigger project that is composed of 2 other final degree thesis. The motivation behind the project was to research ways of solving the challenges that the GNN field currently faces. With GNNs, there is not a single model that can work with all types of graphs, but there exists different strategies that are tuned for every case. Also, the time and memory complexity of a GNN is greatly dependent on the type, size and topology of the graph it is applied to. This can slow down development of GNNs or make it difficult for newcomers to the field to design and deploy their algorithms. The first thesis of this project is aimed to research the characterisation of diverse graph datasets, relevant to GNNs, in order to better understand its properties. The objective of the second thesis is to research a series of algorithms that allows for the acceleration of GNNs and a more efficient memory usage. Finally, this thesis is aimed to propose a way to solve the current challenges in the form of a software tool that can benefit from the knowledge obtained in the other thesis. Named iGNNspector, this tool is intended to explore a way of solving these issues. It is aimed to be used by a wide range of users regarding their experience in GNNs and consists of a framework that also includes a user interface.
MatèriesNeural networks (Computer science), Machine learning, Graph theory, Xarxes neuronals (Informàtica), Aprenentatge automàtic, Grafs, Teoria de
TitulacióGRAU EN ENGINYERIA INFORMÀTICA (Pla 2010)
Col·leccions
Fitxers | Descripció | Mida | Format | Visualitza |
---|---|---|---|---|
159788.pdf | 2,183Mb | Visualitza/Obre |