Generalized Zero-Shot Learning through Similarity Distribution Matching

dc.contributor.advisorShrivastava, Anshumalien_US
dc.creatorDaghaghi, Shabnamen_US
dc.date.accessioned2022-10-05T21:44:55Zen_US
dc.date.available2022-10-05T21:44:55Zen_US
dc.date.created2022-05en_US
dc.date.issued2022-02-18en_US
dc.date.submittedMay 2022en_US
dc.date.updated2022-10-05T21:44:56Zen_US
dc.description.abstractRecent advances in supervised learning methods in vision, specifically deep learning frameworks, are primarily built on the abundance of labeled images. However, image labeling is a laborious task, therefore many visual categories are unlabeled or even unavailable. Zero-Shot Learning (ZSL) deals with classifying unseen visual categories that have no samples during training phase. In particular, ZSL is a classification task where some classes referred to as unseen classes have no training images. Instead, we only have side information about seen and unseen classes, often in the form of semantic or descriptive attributes. Lack of training images from a set of classes restricts the use of standard classification techniques and losses, including the widespread cross-entropy loss. We introduce a novel Similarity Distribution Matching Network (SDM-Net) which is a standard fully connected neural-network architecture with non-trainable penultimate layer consisting of class attributes. The output layer of SDM-Net consists of both seen and unseen classes. To enable zero-shot learning, during training, we regularize the model such that the predicted distribution of unseen class is close in KL divergence to the distribution of similarities between the correct seen class and all the unseen classes. We evaluate the proposed model on five benchmark datasets for zero-shot learning, AwA1, AwA2, aPY, SUN and CUB datasets. We show that, despite the simplicity, our approach achieves competitive performance with state-of-the-art methods in Generalized-ZSL setting for all of these datasets.en_US
dc.format.mimetypeapplication/pdfen_US
dc.identifier.citationDaghaghi, Shabnam. "Generalized Zero-Shot Learning through Similarity Distribution Matching." (2022) Master’s Thesis, Rice University. <a href="https://hdl.handle.net/1911/113538">https://hdl.handle.net/1911/113538</a>.en_US
dc.identifier.urihttps://hdl.handle.net/1911/113538en_US
dc.language.isoengen_US
dc.rightsCopyright is held by the author, unless otherwise indicated. Permission to reuse, publish, or reproduce the work beyond the bounds of fair use or other exemptions to copyright law must be obtained from the copyright holder.en_US
dc.subjectGeneralized Zero-Shot Learningen_US
dc.subjectImage Classificationen_US
dc.subjectSemantic Similarityen_US
dc.subjectKL Divergenceen_US
dc.titleGeneralized Zero-Shot Learning through Similarity Distribution Matchingen_US
dc.typeThesisen_US
dc.type.materialTexten_US
thesis.degree.departmentElectrical and Computer Engineeringen_US
thesis.degree.disciplineEngineeringen_US
thesis.degree.grantorRice Universityen_US
thesis.degree.levelMastersen_US
thesis.degree.nameMaster of Scienceen_US
Files
Original bundle
Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
DAGHAGHI-DOCUMENT-2022.pdf
Size:
2.15 MB
Format:
Adobe Portable Document Format
License bundle
Now showing 1 - 2 of 2
No Thumbnail Available
Name:
PROQUEST_LICENSE.txt
Size:
5.85 KB
Format:
Plain Text
Description:
No Thumbnail Available
Name:
LICENSE.txt
Size:
2.61 KB
Format:
Plain Text
Description: