Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor | Department of Computing | en_US |
| dc.creator | Yau, Yuen Kwan | - |
| dc.identifier.uri | https://theses.lib.polyu.edu.hk/handle/200/11367 | - |
| dc.language | English | en_US |
| dc.publisher | Hong Kong Polytechnic University | en_US |
| dc.rights | All rights reserved | en_US |
| dc.title | Multi-label classification in few-shot context | en_US |
| dcterms.abstract | Few-shot learning is one important emerging method to train machine learning models to learn with only few samples per class. The Prototypical Network model has proven that few-shot learning is a plausible learning method particularly in image classification. However, most of the current researches for few-shot learning focus on classifying an object for one label class at a time. In this work, a model based on prototypical networks for multi-label classifications is proposed. Prototypical Network is a metric-based few shot learning model that learns in the metric space by calculating distances between the prototypes of each class [1]. The target of a prototypical network is to train a model that transforms the raw data like photos and pictures to have distances of prototypes within each class being closer than those between classes in an episode manner. As the number of training samples for each episode is small, the model mimics the few shot learning scenario and may also work with a limited set of data. In our multi-label few shot learning model, we first convert the labels into binary vectors by assigning 0s and 1s for whether the sample has the specific label. During training, distances for samples for each label are taken from the vector means for both 0s and 1s data and a classifier is trained to found am embedding where 0s and 1s images in each label are clustered. In this work, some experiments were successfully carried out and preliminary results for using prototypical networks in multi-label few-shot contest are presented. | en_US |
| dcterms.extent | vii, 47 pages : color illustrations | en_US |
| dcterms.isPartOf | PolyU Electronic Theses | en_US |
| dcterms.issued | 2020 | en_US |
| dcterms.educationalLevel | M.Sc. | en_US |
| dcterms.educationalLevel | All Master | en_US |
| dcterms.LCSH | Machine learning | en_US |
| dcterms.LCSH | Image processing -- Digital techniques | en_US |
| dcterms.LCSH | Hong Kong Polytechnic University -- Dissertations | en_US |
| dcterms.accessRights | restricted access | en_US |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 5815.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 1.21 MB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/11367