Distribution-Based Invariant Deep Networks for Learning Meta-Features
This work addresses the challenge of dataset characterization for machine learning practitioners, offering a novel architecture that improves meta-feature learning, though it is incremental in extending existing invariance concepts.
The paper tackles the problem of learning meta-features from datasets by extending deep learning architectures to be invariant under permutation of both samples and features, proposing Dida. It demonstrates that Dida outperforms state-of-the-art methods on tasks like predicting if dataset patches come from the same source and comparing hyper-parameter performance, achieving better results than DSS, Dataset2Vec, and hand-crafted meta-features.
Recent advances in deep learning from probability distributions successfully achieve classification or regression from distribution samples, thus invariant under permutation of the samples. The first contribution of the paper is to extend these neural architectures to achieve invariance under permutation of the features, too. The proposed architecture, called Dida, inherits the NN properties of universal approximation, and its robustness w.r.t. Lipschitz-bounded transformations of the input distribution is established. The second contribution is to empirically and comparatively demonstrate the merits of the approach on two tasks defined at the dataset level. On both tasks, Dida learns meta-features supporting the characterization of a (labelled) dataset. The first task consists of predicting whether two dataset patches are extracted from the same initial dataset. The second task consists of predicting whether the learning performance achieved by a hyper-parameter configuration under a fixed algorithm (ranging in k-NN, SVM, logistic regression and linear classifier with SGD) dominates that of another configuration, for a dataset extracted from the OpenML benchmarking suite. On both tasks, Dida outperforms the state of the art: DSS (Maron et al., 2020) and Dataset2Vec (Jomaa et al., 2019) architectures, as well as the models based on the hand-crafted meta-features of the literature.