手法を比較
選択した手法を並べて確認できます。異なる行はハイライト表示されます。
| 自己教師あり畳み込みニューラルネットワーク× | 畳み込みニューラルネットワークを用いた転移学習× | |
|---|---|---|
| 分野 | 深層学習 | 深層学習 |
| 系統 | Machine learning | Machine learning |
| 提唱年≠ | 2018–2020 | 2010–2014 |
| 提唱者≠ | LeCun, Y. (CNN backbone); Chen et al. and He et al. (self-supervised visual frameworks) | Pan, S. J. & Yang, Q. (transfer learning framework); popularized for CNNs by Yosinski et al. and Razavian et al. |
| 種類≠ | Self-supervised deep learning | Transfer learning applied to convolutional neural networks |
| 原典≠ | Chen, T., Kornblith, S., Norouzi, M., & Hinton, G. (2020). A Simple Framework for Contrastive Learning of Visual Representations. In Proceedings of the 37th International Conference on Machine Learning (ICML 2020), PMLR 119, 1597–1607. link ↗ | Pan, S. J., & Yang, Q. (2010). A Survey on Transfer Learning. IEEE Transactions on Knowledge and Data Engineering, 22(10), 1345–1359. DOI ↗ |
| 別名 | Self-supervised CNN, SSL-CNN, contrastive CNN, pretext-task CNN | TL-CNN, pretrained CNN, CNN fine-tuning, feature-extracting CNN |
| 関連≠ | 5 | 4 |
| 概要≠ | A self-supervised convolutional neural network (CNN) learns powerful visual representations from unlabeled images by solving pretext tasks — such as contrastive instance discrimination or masked-patch prediction — and then fine-tunes on a small labeled set. This approach dramatically reduces dependence on large annotated datasets while preserving the spatial feature-extraction strengths of convolutional architectures. | Transfer Learning with CNN reuses a convolutional neural network that has already been trained on a large dataset — most commonly ImageNet — and adapts its learned feature detectors to a new, often smaller target dataset. This lets researchers achieve strong image-recognition performance without the massive compute and data resources required to train a CNN from scratch. |
| ScholarGateデータセット ↗ |
|
|