Condensed Data Expansion Using Model Inversion for Knowledge Distillation
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908639311167488 |
|---|---|
| author | Binici, Kuluhan Aggarwal, Shivam Acar, Cihan Pham, Nam Trung Leman, Karianto Lee, Gim Hee Mitra, Tulika |
| author_facet | Binici, Kuluhan Aggarwal, Shivam Acar, Cihan Pham, Nam Trung Leman, Karianto Lee, Gim Hee Mitra, Tulika |
| contents | Condensed datasets offer a compact representation of larger datasets, but training models directly on them or using them to enhance model performance through knowledge distillation (KD) can result in suboptimal outcomes due to limited information. To address this, we propose a method that expands condensed datasets using model inversion, a technique for generating synthetic data based on the impressions of a pre-trained model on its training data. This approach is particularly well-suited for KD scenarios, as the teacher model is already pre-trained and retains knowledge of the original training data. By creating synthetic data that complements the condensed samples, we enrich the training set and better approximate the underlying data distribution, leading to improvements in student model accuracy during knowledge distillation. Our method demonstrates significant gains in KD accuracy compared to using condensed datasets alone and outperforms standard model inversion-based KD methods by up to 11.4% across various datasets and model architectures. Importantly, it remains effective even when using as few as one condensed sample per class, and can also enhance performance in few-shot scenarios where only limited real data samples are available. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_13850 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Condensed Data Expansion Using Model Inversion for Knowledge Distillation Binici, Kuluhan Aggarwal, Shivam Acar, Cihan Pham, Nam Trung Leman, Karianto Lee, Gim Hee Mitra, Tulika Machine Learning Artificial Intelligence Condensed datasets offer a compact representation of larger datasets, but training models directly on them or using them to enhance model performance through knowledge distillation (KD) can result in suboptimal outcomes due to limited information. To address this, we propose a method that expands condensed datasets using model inversion, a technique for generating synthetic data based on the impressions of a pre-trained model on its training data. This approach is particularly well-suited for KD scenarios, as the teacher model is already pre-trained and retains knowledge of the original training data. By creating synthetic data that complements the condensed samples, we enrich the training set and better approximate the underlying data distribution, leading to improvements in student model accuracy during knowledge distillation. Our method demonstrates significant gains in KD accuracy compared to using condensed datasets alone and outperforms standard model inversion-based KD methods by up to 11.4% across various datasets and model architectures. Importantly, it remains effective even when using as few as one condensed sample per class, and can also enhance performance in few-shot scenarios where only limited real data samples are available. |
| title | Condensed Data Expansion Using Model Inversion for Knowledge Distillation |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2408.13850 |