liu.seSearch for publications in DiVA
Change search
ReferencesLink to record
Permanent link

Direct link
Supervised Classification Leveraging Refined Unlabeled Data
Linköping University, Department of Computer and Information Science. Linköping University, Faculty of Science & Engineering.
2015 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

This thesis focuses on how unlabeled data can improve supervised learning classi-fiers in all contexts, for both scarce to abundant label situations. This is meant toaddress the limitations within supervised learning with regards to label availability.Extending the training set with unlabeled data can overcome issues such as selec-tion bias, noise and insufficient data. Based on the overall data distribution andthe initial set of labels, semi-supervised methods provide labels for additional datapoints. The semi-supervised approaches considered in this thesis belong to one ofthe following categories: transductive SVMs, Cluster-then-Label and graph-basedtechniques. Further, we evaluate the behavior of: Logistic regression, Single layerperceptron, SVM and Decision trees. By learning on the extended training set,supervised classifiers are able to generalize better. Based on the results, this the-sis recommends data-processing and algorithmic solutions appropriate to real-worldsituations.

Place, publisher, year, edition, pages
2015. , 64 p.
Keyword [en]
Semi-supervised learning, supervised, active learning, unlabeled, TSVM, graph-based
National Category
Computer and Information Science
URN: urn:nbn:se:liu:diva-119320ISRN: LIU-IDA/STAT-A--15/003—SEOAI: diva2:820890
External cooperation
Subject / course
2015-06-03, 17:58 (English)
Available from: 2015-07-02 Created: 2015-06-12 Last updated: 2015-07-02Bibliographically approved

Open Access in DiVA

fulltext(364 kB)90 downloads
File information
File name FULLTEXT01.pdfFile size 364 kBChecksum SHA-512
Type fulltextMimetype application/pdf

Search in DiVA

By author/editor
Bocancea, Andreea
By organisation
Department of Computer and Information ScienceFaculty of Science & Engineering
Computer and Information Science

Search outside of DiVA

GoogleGoogle Scholar
Total: 90 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Total: 119 hits
ReferencesLink to record
Permanent link

Direct link