Optimized Generic Feature Learning for Few-shot Classification across Domains
This addresses the challenge of generalization in machine learning for applications requiring adaptability across diverse domains, though it appears incremental as it builds on existing HPO methods.
The paper tackles the problem of learning features that generalize across tasks and domains by using cross-domain, cross-task data as a validation objective for hyper-parameter optimization, resulting in learned features that outperform all previous few-shot and meta-learning approaches in few-shot image classification.
To learn models or features that generalize across tasks and domains is one of the grand goals of machine learning. In this paper, we propose to use cross-domain, cross-task data as validation objective for hyper-parameter optimization (HPO) to improve on this goal. Given a rich enough search space, optimization of hyper-parameters learn features that maximize validation performance and, due to the objective, generalize across tasks and domains. We demonstrate the effectiveness of this strategy on few-shot image classification within and across domains. The learned features outperform all previous few-shot and meta-learning approaches.