Self-paced regularization in label distribution learning
Label Distribution Learning is a learning paradigm which outputs a representation of how much each label describes the instance. Research into the paradigm involved machine learning algorithms but did not include deep learning as a possible alternative. Deep learning, a sub-discipline of machine lea...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project (FYP) |
| Language: | English |
| Published: |
2019
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/76943 |
| _version_ | 1826127072714555392 |
|---|---|
| author | Koh, Terence Kang Wei |
| author2 | Bo An |
| author_facet | Bo An Koh, Terence Kang Wei |
| author_sort | Koh, Terence Kang Wei |
| collection | NTU |
| description | Label Distribution Learning is a learning paradigm which outputs a representation of how much each label describes the instance. Research into the paradigm involved machine learning algorithms but did not include deep learning as a possible alternative. Deep learning, a sub-discipline of machine learning, has seen a surge in popularity over the years. However, the process of training is time-consuming as it requires repetitively iterating over a huge training set. To combat this problem, we propose a combinatory method of self-paced regularization with deep learning, where the deep learning model is presented with training data with progressive levels of size, and by extension, difficulty. Experiment results were logged and compared to the traditional deep learning algorithm, as well as state-of-the-art machine learning algorithms. |
| first_indexed | 2024-10-01T07:02:48Z |
| format | Final Year Project (FYP) |
| id | ntu-10356/76943 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T07:02:48Z |
| publishDate | 2019 |
| record_format | dspace |
| spelling | ntu-10356/769432023-03-03T20:31:27Z Self-paced regularization in label distribution learning Koh, Terence Kang Wei Bo An School of Computer Science and Engineering DRNTU::Engineering::Computer science and engineering Label Distribution Learning is a learning paradigm which outputs a representation of how much each label describes the instance. Research into the paradigm involved machine learning algorithms but did not include deep learning as a possible alternative. Deep learning, a sub-discipline of machine learning, has seen a surge in popularity over the years. However, the process of training is time-consuming as it requires repetitively iterating over a huge training set. To combat this problem, we propose a combinatory method of self-paced regularization with deep learning, where the deep learning model is presented with training data with progressive levels of size, and by extension, difficulty. Experiment results were logged and compared to the traditional deep learning algorithm, as well as state-of-the-art machine learning algorithms. Bachelor of Engineering (Computer Science) 2019-04-24T14:48:57Z 2019-04-24T14:48:57Z 2019 Final Year Project (FYP) http://hdl.handle.net/10356/76943 en Nanyang Technological University 29 p. application/pdf |
| spellingShingle | DRNTU::Engineering::Computer science and engineering Koh, Terence Kang Wei Self-paced regularization in label distribution learning |
| title | Self-paced regularization in label distribution learning |
| title_full | Self-paced regularization in label distribution learning |
| title_fullStr | Self-paced regularization in label distribution learning |
| title_full_unstemmed | Self-paced regularization in label distribution learning |
| title_short | Self-paced regularization in label distribution learning |
| title_sort | self paced regularization in label distribution learning |
| topic | DRNTU::Engineering::Computer science and engineering |
| url | http://hdl.handle.net/10356/76943 |
| work_keys_str_mv | AT kohterencekangwei selfpacedregularizationinlabeldistributionlearning |