Unsupervised Meta-Learning through Latent-Space Interpolation in Generative Models

Siavash Khodadadeh, Sharare Zehtabian, Saeed Vahidian, Weijia Wang, Bill Lin, Ladislau Boloni

Keywords: Meta-learning, Unsupervised learning, GANs

Abstract: Several recently proposed unsupervised meta-learning approaches rely on synthetic meta-tasks created using techniques such as random selection, clustering and/or augmentation. In this work, we describe a novel approach that generates meta-tasks using generative models. The proposed family of algorithms generate pairs of in-class and out-of-class samples from the latent space in a principled way, allowing us to create synthetic classes forming the training and validation data of a meta-task. We find that the proposed approach, LAtent Space Interpolation Unsupervised Meta-learning (LASIUM), outperforms or is competitive with current unsupervised learning baselines on few-shot classification tasks on the most widely used benchmark datasets.

One-sentence Summary: We use interpolation in generative models latent space to generate tasks for unsupervised meta-learninig.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Data: [CelebA](https://paperswithcode.com/dataset/celeba), [mini-Imagenet](https://paperswithcode.com/dataset/mini-imagenet)