Repository logo
 

Evolving Hierarchical Structures for Convolutional Neural Networks using Jagged Arrays

dc.contributor.authorMcIlroy, Stuart
dc.contributor.copyright-releaseNot Applicableen_US
dc.contributor.degreeMaster of Computer Scienceen_US
dc.contributor.departmentFaculty of Computer Scienceen_US
dc.contributor.ethics-approvalNot Applicableen_US
dc.contributor.external-examinern/aen_US
dc.contributor.graduate-coordinatorNorbert Zehen_US
dc.contributor.manuscriptsNot Applicableen_US
dc.contributor.thesis-readerMalcolm Heywooden_US
dc.contributor.thesis-readerSageev Ooreen_US
dc.contributor.thesis-supervisorThomas Trappenbergen_US
dc.date.accessioned2018-04-04T17:47:19Z
dc.date.available2018-04-04T17:47:19Z
dc.date.defence2018-03-28
dc.date.issued2018-04-04T17:47:19Z
dc.description.abstractTraditionally, deep learning practitioners have relied on heuristics and past high-performing networks to hand-craft their neural network architectures because architecture search algorithms were too computationally intensive. However, with the advent of better GPU processing for neural networks, architecture search is now feasible. We propose a search through the architecture space, using a genetic algorithm that evolves convolutional neural networks using a novel decomposition scheme of nested hierarchical structures using jagged arrays. We test it on standard image classification benchmarks (MNIST, CIFAR-10) as well as a modular task (classifying "what" and "where"). The resulting architectures achieve performance comparable with the state of the art. We then explore the types of network architectures that arise from the evolutionary stage. We show that the evolved architectures adapt to the specific dataset, that common heuristics such as modular reuse evolve independently in our approach and hierarchical structures develop for the modular task.en_US
dc.identifier.urihttp://hdl.handle.net/10222/73819
dc.language.isoenen_US
dc.subjectdeep learningen_US
dc.subjectarchitecture searchen_US
dc.subjectgenetic algorithmen_US
dc.titleEvolving Hierarchical Structures for Convolutional Neural Networks using Jagged Arraysen_US

Files

Original bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
Mcilroy-Stuart-MSc-CSCI-March-2018.pdf
Size:
1.81 MB
Format:
Adobe Portable Document Format
Description:

License bundle

Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
1.71 KB
Format:
Item-specific license agreed upon to submission
Description: