Inlämning av Examensarbete / Submission of Thesis

Fredrik Åström , pp. 30. COM/School of Computing, 2011.

The work

Författare / Author: Fredrik Åström
fredrik.astrom89@gmail.com
Titel / Title: Neural Network on Compute Shader - Running and Training a Neural Network using GPGPU
Abstrakt Abstract:

In this thesis I look into how one can train and run an artificial neural network using Compute Shader and what kind of performance can be expected.

An artificial neural network is a computational model that is inspired by biological neural networks, e.g. a brain.

Finding what kind of performance can be expected was done by creating an implementation that uses Compute Shader and then compare it to the FANN library, i.e. a fast artificial neural network library written in C.

The conclusion is that you can improve performance by training an artificial neural network on the compute shader as long as you are using non-trivial datasets and neural network configurations.

Ämnesord / Subject: Datavetenskap - Computer Science\Artificial Intelligence
Datavetenskap - Computer Science\General
Nyckelord / Keywords: Artificial Neural Network, GPGPU, Compute Shader

Publication info

Dokument id / Document id: houn-8kd3jp
Program:/ Programme Spelprogrammering/Digital Game Development
Registreringsdatum / Date of registration: 07/06/2011
Uppsatstyp / Type of thesis: C-Uppsats

Context

Handledare / Supervisor: Stefan Johansson
sja@bth.se
Examinator / Examiner: Stefan Johansson
Organisation / Organisation: Blekinge Institute of Technology
Institution / School: COM/School of Computing

+46 455 38 50 00

Files & Access

Bifogad uppsats fil(er) / Files attached: bth2011astrom.pdf (573 kB, öppnas i nytt fönster)