Leaf shape recognition via support vector machines with edit distance kernels Public Deposited

http://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/gq67jt19z

Descriptions

Attribute NameValues
Creator
Abstract or Summary
  • Edit distances are a well-established technique for classification problems. They have been employed successfully in many classification problems including chromosome classification and hand-written digit recognition. Virtually all machine learning algorithms represent the objects to be classified as vectors of features. However, edit distances provide only a measure of the difference between two objects--they do not provide a feature-based representation. Recently, kernel-based algorithms such as support vector machines (SVMs) have been developed. A kernel is a measure of similarity between two objects. This thesis explores various ways in which edit distances can be converted into kernels and combined with support vector machines to solve difficult pattern recognition problems. The thesis compares the performance of SVMs, the k-nearest neighbor algorithm, and the weighted k-nearest neighbor algorithm on a problem of leaf shape classification. The goal of the leaf classification problem is support content-based image retrieval from the Oregon State University Herbarium, which is a collection of plant specimens. The thesis shows that SVMs and standard k-nearest neighbor provide high classification accuracy. SVMs, with an edit distance kernel, provide the most accurate method.
Resource Type
Date Available
Date Copyright
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Academic Affiliation
Non-Academic Affiliation
Subject
Rights Statement
Language
Digitization Specifications
  • PDF derivative scanned at 300 ppi (24-bit Color and 256 B&W), using Capture Perfect 3.0.82, on a Canon DR-9080C. CVista PdfCompressor 4.0 was used for pdf compression and textual OCR.
Replaces
Additional Information
  • description.provenance : Approved for entry into archive by Linda Kathman(linda.kathman@oregonstate.edu) on 2009-06-09T14:33:25Z (GMT) No. of bitstreams: 1 Sinha_Shriprakash_2004.pdf: 729171 bytes, checksum: 6671ee8206c75a2aa9f7bbc338ddc743 (MD5)
  • description.provenance : Submitted by Anna Opoien (aoscanner@gmail.com) on 2009-06-08T19:21:33Z No. of bitstreams: 1 Sinha_Shriprakash_2004.pdf: 729171 bytes, checksum: 6671ee8206c75a2aa9f7bbc338ddc743 (MD5)
  • description.provenance : Made available in DSpace on 2009-06-09T14:35:22Z (GMT). No. of bitstreams: 1 Sinha_Shriprakash_2004.pdf: 729171 bytes, checksum: 6671ee8206c75a2aa9f7bbc338ddc743 (MD5)
  • description.provenance : Approved for entry into archive by Linda Kathman(linda.kathman@oregonstate.edu) on 2009-06-09T14:35:22Z (GMT) No. of bitstreams: 1 Sinha_Shriprakash_2004.pdf: 729171 bytes, checksum: 6671ee8206c75a2aa9f7bbc338ddc743 (MD5)

Relationships

In Administrative Set:
Last modified: 08/01/2017

Downloadable Content

Download PDF
Citations:

EndNote | Zotero | Mendeley

Items