Maximum Entropy Fine-Grained Classification
© 2018 Curran Associates Inc..All rights reserved. Fine-Grained Visual Classification (FGVC) is an important computer vision problem that involves small diversity within the different classes, and often requires expert annotators to collect data. Utilizing this notion of small visual diversity, we r...
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
2021
|
| Online Access: | https://hdl.handle.net/1721.1/138084 |
| _version_ | 1826197058924576768 |
|---|---|
| author | Dubey, Abhimanyu Gupta, Otkrist Raskar, Ramesh Naik, Nikhil |
| author2 | Massachusetts Institute of Technology. Media Laboratory |
| author_facet | Massachusetts Institute of Technology. Media Laboratory Dubey, Abhimanyu Gupta, Otkrist Raskar, Ramesh Naik, Nikhil |
| author_sort | Dubey, Abhimanyu |
| collection | MIT |
| description | © 2018 Curran Associates Inc..All rights reserved. Fine-Grained Visual Classification (FGVC) is an important computer vision problem that involves small diversity within the different classes, and often requires expert annotators to collect data. Utilizing this notion of small visual diversity, we revisit Maximum-Entropy learning in the context of fine-grained classification, and provide a training routine that maximizes the entropy of the output probability distribution for training convolutional neural networks on FGVC tasks. We provide a theoretical as well as empirical justification of our approach, and achieve state-of-the-art performance across a variety of classification tasks in FGVC, that can potentially be extended to any fine-tuning task. Our method is robust to different hyperparameter values, amount of training data and amount of training label noise and can hence be a valuable tool in many similar problems. |
| first_indexed | 2024-09-23T10:42:11Z |
| format | Article |
| id | mit-1721.1/138084 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T10:42:11Z |
| publishDate | 2021 |
| record_format | dspace |
| spelling | mit-1721.1/1380842023-02-09T19:17:31Z Maximum Entropy Fine-Grained Classification Dubey, Abhimanyu Gupta, Otkrist Raskar, Ramesh Naik, Nikhil Massachusetts Institute of Technology. Media Laboratory © 2018 Curran Associates Inc..All rights reserved. Fine-Grained Visual Classification (FGVC) is an important computer vision problem that involves small diversity within the different classes, and often requires expert annotators to collect data. Utilizing this notion of small visual diversity, we revisit Maximum-Entropy learning in the context of fine-grained classification, and provide a training routine that maximizes the entropy of the output probability distribution for training convolutional neural networks on FGVC tasks. We provide a theoretical as well as empirical justification of our approach, and achieve state-of-the-art performance across a variety of classification tasks in FGVC, that can potentially be extended to any fine-tuning task. Our method is robust to different hyperparameter values, amount of training data and amount of training label noise and can hence be a valuable tool in many similar problems. 2021-11-09T21:44:50Z 2021-11-09T21:44:50Z 2018 2019-08-02T14:43:20Z Article http://purl.org/eprint/type/ConferencePaper https://hdl.handle.net/1721.1/138084 Dubey, Abhimanyu, Gupta, Otkrist, Raskar, Ramesh and Naik, Nikhil. 2018. "Maximum Entropy Fine-Grained Classification." en Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. application/pdf Neural Information Processing Systems (NIPS) |
| spellingShingle | Dubey, Abhimanyu Gupta, Otkrist Raskar, Ramesh Naik, Nikhil Maximum Entropy Fine-Grained Classification |
| title | Maximum Entropy Fine-Grained Classification |
| title_full | Maximum Entropy Fine-Grained Classification |
| title_fullStr | Maximum Entropy Fine-Grained Classification |
| title_full_unstemmed | Maximum Entropy Fine-Grained Classification |
| title_short | Maximum Entropy Fine-Grained Classification |
| title_sort | maximum entropy fine grained classification |
| url | https://hdl.handle.net/1721.1/138084 |
| work_keys_str_mv | AT dubeyabhimanyu maximumentropyfinegrainedclassification AT guptaotkrist maximumentropyfinegrainedclassification AT raskarramesh maximumentropyfinegrainedclassification AT naiknikhil maximumentropyfinegrainedclassification |