A Meta-Learning Approach for Custom Model Training
This work addresses a limitation in meta-learning for custom model training, offering a more robust solution for practitioners dealing with diverse task complexities, though it is incremental as it builds on existing transfer and meta-learning methods.
The paper tackles the problem of meta-learning algorithms not generalizing well from few-class, few-shot settings to many-class, many-shot scenarios by proposing a joint training approach that combines transfer-learning and meta-learning, resulting in improved generalization performance on unseen target tasks across both settings.
Transfer-learning and meta-learning are two effective methods to apply knowledge learned from large data sources to new tasks. In few-class, few-shot target task settings (i.e. when there are only a few classes and training examples available in the target task), meta-learning approaches that optimize for future task learning have outperformed the typical transfer approach of initializing model weights from a pre-trained starting point. But as we experimentally show, meta-learning algorithms that work well in the few-class setting do not generalize well in many-shot and many-class cases. In this paper, we propose a joint training approach that combines both transfer-learning and meta-learning. Benefiting from the advantages of each, our method obtains improved generalization performance on unseen target tasks in both few- and many-class and few- and many-shot scenarios.