Sieve: Multimodal Dataset Pruning Using Image Captioning Models
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2023
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866917608883748864 |
|---|---|
| author | Mahmoud, Anas Elhoushi, Mostafa Abbas, Amro Yang, Yu Ardalani, Newsha Leather, Hugh Morcos, Ari |
| author_facet | Mahmoud, Anas Elhoushi, Mostafa Abbas, Amro Yang, Yu Ardalani, Newsha Leather, Hugh Morcos, Ari |
| contents | Vision-Language Models (VLMs) are pretrained on large, diverse, and noisy web-crawled datasets. This underscores the critical need for dataset pruning, as the quality of these datasets is strongly correlated with the performance of VLMs on downstream tasks. Using CLIPScore from a pretrained model to only train models using highly-aligned samples is one of the most successful methods for pruning. We argue that this approach suffers from multiple limitations including: false positives and negatives due to CLIP's pretraining on noisy labels. We propose a pruning signal, Sieve, that employs synthetic captions generated by image-captioning models pretrained on small, diverse, and well-aligned image-text pairs to evaluate the alignment of noisy image-text pairs. To bridge the gap between the limited diversity of generated captions and the high diversity of alternative text (alt-text), we estimate the semantic textual similarity in the embedding space of a language model pretrained on unlabeled text corpus. Using DataComp, a multimodal dataset filtering benchmark, when evaluating on 38 downstream tasks, our pruning approach, surpasses CLIPScore by 2.6\% and 1.7\% on medium and large scale respectively. In addition, on retrieval tasks, Sieve leads to a significant improvement of 2.7% and 4.5% on medium and large scale respectively. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2310_02110 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Sieve: Multimodal Dataset Pruning Using Image Captioning Models Mahmoud, Anas Elhoushi, Mostafa Abbas, Amro Yang, Yu Ardalani, Newsha Leather, Hugh Morcos, Ari Computer Vision and Pattern Recognition Vision-Language Models (VLMs) are pretrained on large, diverse, and noisy web-crawled datasets. This underscores the critical need for dataset pruning, as the quality of these datasets is strongly correlated with the performance of VLMs on downstream tasks. Using CLIPScore from a pretrained model to only train models using highly-aligned samples is one of the most successful methods for pruning. We argue that this approach suffers from multiple limitations including: false positives and negatives due to CLIP's pretraining on noisy labels. We propose a pruning signal, Sieve, that employs synthetic captions generated by image-captioning models pretrained on small, diverse, and well-aligned image-text pairs to evaluate the alignment of noisy image-text pairs. To bridge the gap between the limited diversity of generated captions and the high diversity of alternative text (alt-text), we estimate the semantic textual similarity in the embedding space of a language model pretrained on unlabeled text corpus. Using DataComp, a multimodal dataset filtering benchmark, when evaluating on 38 downstream tasks, our pruning approach, surpasses CLIPScore by 2.6\% and 1.7\% on medium and large scale respectively. In addition, on retrieval tasks, Sieve leads to a significant improvement of 2.7% and 4.5% on medium and large scale respectively. |
| title | Sieve: Multimodal Dataset Pruning Using Image Captioning Models |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2310.02110 |