방법 비교
선택한 방법을 나란히 검토하세요. 서로 다른 행은 강조 표시됩니다.
| 준지도 학습 이미지 분류× | 미세 조정된 이미지 분류× | |
|---|---|---|
| 분야 | 딥러닝 | 딥러닝 |
| 계열 | Machine learning | Machine learning |
| 기원 연도≠ | 2013–2020 | 2010–2014 |
| 창시자≠ | Lee, D.-H. (pseudo-label); Sohn et al. (FixMatch) | Yosinski, J. et al.; Pan, S. J. & Yang, Q. |
| 유형≠ | Semi-supervised deep learning | Transfer learning / fine-tuning |
| 원전≠ | Lee, D.-H. (2013). Pseudo-Label: The Simple and Efficient Semi-Supervised Learning Method for Deep Neural Networks. ICML 2013 Workshop on Challenges in Representation Learning. link ↗ | Yosinski, J., Clune, J., Bengio, Y., & Lipson, H. (2014). How transferable are features in deep neural networks? Advances in Neural Information Processing Systems (NeurIPS), 27, 3320–3328. link ↗ |
| 별칭 | SSL image classification, semi-supervised CNN classification, pseudo-label image classification, label-efficient image classification | fine-tuning for image recognition, transfer learning image classifier, pretrained CNN fine-tuning, domain-specific image classifier |
| 관련 | 5 | 5 |
| 요약≠ | Semi-supervised image classification trains deep neural networks on a small set of labeled images together with a much larger pool of unlabeled images. Techniques such as pseudo-labeling, consistency regularization, and confidence thresholding allow the model to leverage the structure of unlabeled data, dramatically reducing the need for expensive manual annotation while approaching fully-supervised accuracy. | Fine-tuned image classification adapts a large neural network pretrained on a broad image corpus (such as ImageNet) to a specific target domain by continuing training on labeled domain images. This approach achieves strong accuracy with far fewer target-domain samples than training from scratch, making it the dominant paradigm for applied computer vision tasks. |
| ScholarGate데이터셋 ↗ |
|
|