Instance-Level Embedding Adaptation for Few-Shot Learning
Few-shot learning aims to recognize novel categories from just a few labeled instances. Existing metric learning-based approaches perform classifications by nearest neighbor search in the embedding space. The embedding function is a deep neural network and usually shared by all novel categories. How...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2019-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/8672561/ |
| _version_ | 1818737308508094464 |
|---|---|
| author | Fusheng Hao Jun Cheng Lei Wang Jianzhong Cao |
| author_facet | Fusheng Hao Jun Cheng Lei Wang Jianzhong Cao |
| author_sort | Fusheng Hao |
| collection | DOAJ |
| description | Few-shot learning aims to recognize novel categories from just a few labeled instances. Existing metric learning-based approaches perform classifications by nearest neighbor search in the embedding space. The embedding function is a deep neural network and usually shared by all novel categories. However, these brute approaches lack a fast adaptation mechanism like meta-learning when dealing with novel categories. To tackle this, we present a novel instance-level embedding adaptation mechanism, aiming at rapidly adapting embedding deep features to improve their generalization ability in recognizing novel categories. To this end, we design an Attention Adaptation Module to pull a query instance and its corresponding class center as close as possible. Note that, each query instance is pulled closer to its corresponding class center before performing nearest neighbor classifications. This instance-level reduction of intra-class distance increases the probability of correct classifications, and thus improves the generalization ability to embed deep features and promoting the performance. The extensive experiments are conducted on two benchmark datasets: miniImageNet and CUB. Our approach yields very promising results on both datasets. In addition, in a realistic cross-domain evaluation setting, our method also achieves the-state-of-the-art performance. |
| first_indexed | 2024-12-18T00:50:59Z |
| format | Article |
| id | doaj.art-63b677d1964149348759f538effa9f34 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-18T00:50:59Z |
| publishDate | 2019-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-63b677d1964149348759f538effa9f342022-12-21T21:26:41ZengIEEEIEEE Access2169-35362019-01-01710050110051110.1109/ACCESS.2019.29066658672561Instance-Level Embedding Adaptation for Few-Shot LearningFusheng Hao0Jun Cheng1https://orcid.org/0000-0002-3131-3275Lei Wang2https://orcid.org/0000-0001-5990-896XJianzhong Cao3Chinese Academy of Sciences, Xi’an Institute of Optics and Precision Mechanics, Xi’an, ChinaShenzhen Key Laboratory of Virtual Reality and Human Interaction Technology, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, ChinaShenzhen Key Laboratory of Virtual Reality and Human Interaction Technology, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, ChinaChinese Academy of Sciences, Xi’an Institute of Optics and Precision Mechanics, Xi’an, ChinaFew-shot learning aims to recognize novel categories from just a few labeled instances. Existing metric learning-based approaches perform classifications by nearest neighbor search in the embedding space. The embedding function is a deep neural network and usually shared by all novel categories. However, these brute approaches lack a fast adaptation mechanism like meta-learning when dealing with novel categories. To tackle this, we present a novel instance-level embedding adaptation mechanism, aiming at rapidly adapting embedding deep features to improve their generalization ability in recognizing novel categories. To this end, we design an Attention Adaptation Module to pull a query instance and its corresponding class center as close as possible. Note that, each query instance is pulled closer to its corresponding class center before performing nearest neighbor classifications. This instance-level reduction of intra-class distance increases the probability of correct classifications, and thus improves the generalization ability to embed deep features and promoting the performance. The extensive experiments are conducted on two benchmark datasets: miniImageNet and CUB. Our approach yields very promising results on both datasets. In addition, in a realistic cross-domain evaluation setting, our method also achieves the-state-of-the-art performance.https://ieeexplore.ieee.org/document/8672561/Attention adaptation moduledeep neural networksfew-shot learning |
| spellingShingle | Fusheng Hao Jun Cheng Lei Wang Jianzhong Cao Instance-Level Embedding Adaptation for Few-Shot Learning IEEE Access Attention adaptation module deep neural networks few-shot learning |
| title | Instance-Level Embedding Adaptation for Few-Shot Learning |
| title_full | Instance-Level Embedding Adaptation for Few-Shot Learning |
| title_fullStr | Instance-Level Embedding Adaptation for Few-Shot Learning |
| title_full_unstemmed | Instance-Level Embedding Adaptation for Few-Shot Learning |
| title_short | Instance-Level Embedding Adaptation for Few-Shot Learning |
| title_sort | instance level embedding adaptation for few shot learning |
| topic | Attention adaptation module deep neural networks few-shot learning |
| url | https://ieeexplore.ieee.org/document/8672561/ |
| work_keys_str_mv | AT fushenghao instancelevelembeddingadaptationforfewshotlearning AT juncheng instancelevelembeddingadaptationforfewshotlearning AT leiwang instancelevelembeddingadaptationforfewshotlearning AT jianzhongcao instancelevelembeddingadaptationforfewshotlearning |