Meta-learning for regularizing deep networks under small data regimes (C05)

我们的目标是扩展在相当大的表格数据集上所取得的进展，以“缩小”深度学习，使其在小型数据集上也有效。具体来说，我们将开发方法来搜索正则化方法的最佳组合，基于跨许多小数据集的元学习方法，并通过集成正则化方法的不同组合。我们还将处理纵向数据和图像数据的更结构化的数据形式。

We aim at extending the advances made for rather large tabular datasets to "scale down” deep learning to be effective also in the regime of small datasets. Specifically, we will develop approaches to search for optimal combinations of regularization methods, based on a meta-learning approach across many small datasets and by ensembling different combinations of regularization methods. We will also tackle the more structured data modalities of longitudinal data and image data.