Порівняння методів
Переглядайте обрані методи поруч; рядки з відмінностями підсвічено.
| Самокерований Гаусів процес× | Активне навчання на Гауссових процесах× | |
|---|---|---|
| Галузь | Машинне навчання | Машинне навчання |
| Родина | Machine learning | Machine learning |
| Рік появи≠ | 2019–2021 | 1992 |
| Автор методу≠ | Fortuin, V. et al.; broader self-supervised GP literature | MacKay, D. J. C. |
| Тип≠ | Probabilistic model (self-supervised GP pretraining + kernel learning) | Bayesian active learning |
| Основоположне джерело≠ | Fortuin, V., Rätsch, G., & Mandt, S. (2020). GP-VAE: Deep probabilistic time series imputation using Gaussian process variational autoencoders. Proceedings of the 23rd International Conference on Artificial Intelligence and Statistics (AISTATS), PMLR 108, 1651–1661. link ↗ | MacKay, D. J. C. (1992). Information-based objective functions for active data selection. Neural Computation, 4(4), 590–604. DOI ↗ |
| Інші назви | SSL-GP, self-supervised GP, self-supervised GPR, self-supervised Gaussian process regression | GP active learning, Gaussian process active learning, GP-AL, Bayesian active learning with GP |
| Пов'язані≠ | 6 | 4 |
| Підсумок≠ | Self-supervised Gaussian Process (SSL-GP) combines the principled uncertainty quantification of Gaussian processes with self-supervised pretraining, learning expressive kernels or latent representations from unlabeled data before fitting a GP on a small labeled set. This makes the approach especially powerful in low-labeled-data regimes where a conventional GP would overfit or produce poorly calibrated uncertainty estimates. | Active Learning Gaussian Process (GP-AL) combines a Gaussian process probabilistic model with an active learning query strategy, using the GP's posterior uncertainty to select the most informative unlabeled examples for labeling. This iterative approach minimizes labeling effort while maximizing predictive accuracy, making it ideal when labeled data is scarce or expensive to obtain. |
| ScholarGateНабір даних ↗ |
|
|