Distribution Embedding Networks for Generalization from a Diverse Set of Classification Tasks

TL;DR

Distribution Embedding Networks (DEN) enable effective classification across diverse and heterogeneous tasks by learning fixed embeddings and minimal task-specific adaptation, excelling especially in small data and tabular settings.

Contribution

DEN introduces a novel architecture that generalizes across heterogeneous tasks and allows fixed embeddings, with a simple fine-tuning step for new tasks, supported by synthetic task synthesis.

Findings

DEN outperforms existing methods on synthetic and real datasets.

The architecture allows fixed embeddings after pre-training, requiring minimal fine-tuning.

Synthetic task generation facilitates training and improves generalization.

Abstract

We propose Distribution Embedding Networks (DEN) for classification with small data. In the same spirit of meta-learning, DEN learns from a diverse set of training tasks with the goal to generalize to unseen target tasks. Unlike existing approaches which require the inputs of training and target tasks to have the same dimension with possibly similar distributions, DEN allows training and target tasks to live in heterogeneous input spaces. This is especially useful for tabular-data tasks where labeled data from related tasks are scarce. DEN uses a three-block architecture: a covariate transformation block followed by a distribution embedding block and then a classification block. We provide theoretical insights to show that this architecture allows the embedding and classification blocks to be fixed after pre-training on a diverse set of tasks; only the covariate transformation block with relatively few parameters needs to be fine-tuned for each new task. To facilitate training, we also propose an approach to synthesize binary classification tasks, and demonstrate that DEN outperforms existing methods in a number of synthetic and real tasks in numerical studies.