Porovnať metódy
Prezrite si vybrané metódy vedľa seba; riadky, ktoré sa líšia, sú zvýraznené.
| Regularizované učenie s malým počtom príkladov (Regularized Few-Shot Learning)× | Samoučenie× | |
|---|---|---|
| Odbor | Strojové učenie | Strojové učenie |
| Rodina | Machine learning | Machine learning |
| Rok vzniku≠ | 2016-2020 | 2018–2020 |
| Tvorca≠ | Multiple (Chen et al., Tian et al., Snell et al., and others) | LeCun, Y. and community (formalized ~2018–2020) |
| Typ≠ | Meta-learning framework with explicit regularization | Representation learning paradigm |
| Pôvodný zdroj≠ | Chen, W., Liu, Y., Kira, Z., Wang, Y. F., & Huang, J. (2019). A Closer Look at Few-Shot Classification. International Conference on Learning Representations (ICLR). link ↗ | LeCun, Y. & Misra, I. (2022). Self-supervised learning: The dark matter of intelligence. Meta AI Blog. https://ai.facebook.com/blog/self-supervised-learning-the-dark-matter-of-intelligence/ link ↗ |
| Ďalšie názvy | FSL with regularization, regularized meta-learning, few-shot learning with regularization, regularized episodic learning | SSL, self-supervised pre-training, pretext-task learning, unsupervised representation learning |
| Príbuzné≠ | 5 | 3 |
| Zhrnutie≠ | Regularized few-shot learning augments standard few-shot learning pipelines with explicit regularization mechanisms — such as weight decay, dropout, data augmentation, label smoothing, or manifold constraints — to reduce overfitting to the tiny support sets that define each episode. This produces more generalizable models when only one to thirty labeled examples per class are available. | Self-supervised learning (SSL) is a machine-learning paradigm that generates its own supervisory signal directly from unlabeled data by defining an auxiliary pretext task — such as predicting masked words, rotating images, or contrasting augmented views — and uses the learned representations as a powerful starting point for downstream tasks with minimal labeled examples. |
| ScholarGateDátová sada ↗ |
|
|