Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2404.00901 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929299132514304 |
|---|---|
| author | Jiao, Jian Dai, Yu Mei, Hefei Qiu, Heqian Gong, Chuanyang Tang, Shiyuan Hao, Xinpeng Li, Hongliang |
| author_facet | Jiao, Jian Dai, Yu Mei, Hefei Qiu, Heqian Gong, Chuanyang Tang, Shiyuan Hao, Xinpeng Li, Hongliang |
| contents | Recent video class-incremental learning usually excessively pursues the accuracy of the newly seen classes and relies on memory sets to mitigate catastrophic forgetting of the old classes. However, limited storage only allows storing a few representative videos. So we propose SNRO, which slightly shifts the features of new classes to remember old classes. Specifically, SNRO contains Examples Sparse(ES) and Early Break(EB). ES decimates at a lower sample rate to build memory sets and uses interpolation to align those sparse frames in the future. By this, SNRO stores more examples under the same memory consumption and forces the model to focus on low-semantic features which are harder to be forgotten. EB terminates the training at a small epoch, preventing the model from overstretching into the high-semantic space of the current task. Experiments on UCF101, HMDB51, and UESTC-MMEA-CL datasets show that SNRO performs better than other approaches while consuming the same memory consumption. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_00901 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Slightly Shift New Classes to Remember Old Classes for Video Class-Incremental Learning Jiao, Jian Dai, Yu Mei, Hefei Qiu, Heqian Gong, Chuanyang Tang, Shiyuan Hao, Xinpeng Li, Hongliang Computer Vision and Pattern Recognition Recent video class-incremental learning usually excessively pursues the accuracy of the newly seen classes and relies on memory sets to mitigate catastrophic forgetting of the old classes. However, limited storage only allows storing a few representative videos. So we propose SNRO, which slightly shifts the features of new classes to remember old classes. Specifically, SNRO contains Examples Sparse(ES) and Early Break(EB). ES decimates at a lower sample rate to build memory sets and uses interpolation to align those sparse frames in the future. By this, SNRO stores more examples under the same memory consumption and forces the model to focus on low-semantic features which are harder to be forgotten. EB terminates the training at a small epoch, preventing the model from overstretching into the high-semantic space of the current task. Experiments on UCF101, HMDB51, and UESTC-MMEA-CL datasets show that SNRO performs better than other approaches while consuming the same memory consumption. |
| title | Slightly Shift New Classes to Remember Old Classes for Video Class-Incremental Learning |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2404.00901 |