Heterogeneous Computing for Data Stream Mining

Simple item page
dc.contributor.advisorPfahringer, Bernhard
dc.contributor.authorPetko, Vladimir
dc.date.accessioned2016-07-12T23:15:43Z
dc.date.available2016-07-12T23:15:43Z
dc.date.issued2016
dc.date.updated2016-03-17T00:01:42Z
dc.description.abstractGraphical Processing Units are de-facto standard for acceleration of data parallel tasks in high performance computing. They are widely used to accelerate batch machine learning algorithms. High-end discrete GPUs are characterized by a very high number of cores (thousands), high bandwidth memory optimized for the stream access and high power requirements. Integrated GPUs are characterized by a medium number of cores (hundreds), medium bandwidth memory shared with CPU optimized for the random access and low power requirements. Data stream processing applications are often required to provide response within the limited time frame, operate on data in relatively small increments and have strict power requirements if deployed on the embedded devices. This work evaluates performance of integrated and discrete GPUs belonging to the same chip family on several variants of k-nearest neighbours algorithm over sliding window and stochastic gradient descent using OpenCL and novel Heterogeneous System Architecture platforms. We conclude that integrated GPUs provide a niche solution catering for to small work sizes that offers better power efficiency and simplicity of deployment.
dc.format.mimetypeapplication/pdf
dc.identifier.citationPetko, V. (2016). Heterogeneous Computing for Data Stream Mining (Thesis, Master of Science (Research) (MSc(Research))). University of Waikato, Hamilton, New Zealand. Retrieved from https://hdl.handle.net/10289/10524en
dc.identifier.urihttps://hdl.handle.net/10289/10524
dc.language.isoen
dc.publisherUniversity of Waikato
dc.rightsAll items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.
dc.subjectdata stream mining
dc.subjectgpu
dc.subjectknn
dc.subjectsgd
dc.subjectz-order
dc.subjectk-d tree
dc.subjectrandom projection tree
dc.titleHeterogeneous Computing for Data Stream Mining
dc.typeThesis
pubs.place-of-publicationHamilton, New Zealanden_NZ
thesis.degree.grantorUniversity of Waikato
thesis.degree.levelMasters
thesis.degree.nameMaster of Science (Research) (MSc(Research))
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
thesis.pdf
Size:
1.29 MB
Format:
Adobe Portable Document Format
Description:
Thumbnail Image
Name: thesis.pdf
Size: 1.29 MB
Kind: Adobe PDF
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
2.07 KB
Format:
Item-specific license agreed upon to submission
Description:
Name: license.txt
Size: 2.07 KB
Kind: License
Collections
Masters Degree Theses