Large-Scale Few-Shot Learning Challenge

Many Categories × Few Examples

The goal of this challenge is to create a large-scale platform for benchmarking few-shot, one-shot, and zero-shot learning algorithms. In contrast to existing challenges with many examples, we focus on increasing the number of categories with high-quality few examples.

Representing Concepts

Our main interest lies in representations of concepts. It includes, but not limited to,
 - Text Represetations
 - Image Representations
 - Video Representations
 - Audio Representations
 - Sensor-Data Representations
and the relation among them.

Few-Shot Verb Image Dataset

Few-Shot Verb Image Classification is our first challenge. We provide high-quality image urls that represent verb synsets in the WordNet. Visit BrowseSearch, DownloadDownload, and ChallengeWorkshop for details.