Porovnat metody

	Polo-dohledná klasifikace obrazu	Dolaďování klasifikace obrazu
Obor	Hluboké učení	Hluboké učení
Rodina	Machine learning	Machine learning
Rok vzniku	2013–2020	2010–2014
Tvůrce	Lee, D.-H. (pseudo-label); Sohn et al. (FixMatch)	Yosinski, J. et al.; Pan, S. J. & Yang, Q.
Typ	Semi-supervised deep learning	Transfer learning / fine-tuning
Původní zdroj	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.	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.
Další názvy	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
Příbuzné	5	5
Shrnutí	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.