POWN: Prototypical Open-World Node Classification

TL;DR

POWN introduces a novel end-to-end method for open-world node classification that effectively distinguishes known and new classes using prototypes, outperforming existing approaches without requiring data augmentation.

Contribution

It presents a new prototype-based approach for open-world node classification that combines multiple learning strategies and does not need data augmentation, advancing the state of the art.

Findings

Outperforms baseline methods by up to 20% accuracy on small datasets.

Achieves up to 30% accuracy improvement on large datasets.

Effectively distinguishes between known and multiple new classes.

Abstract

We consider the problem of \textit{true} open-world semi-supervised node classification, in which nodes in a graph either belong to known or new classes, with the latter not present during training. Existing methods detect and reject new classes but fail to distinguish between different new classes. We adapt existing methods and show they do not solve the problem sufficiently. We introduce a novel end-to-end approach for classification into known classes and new classes based on class prototypes, which we call Prototypical Open-World Learning for Node Classification (POWN). Our method combines graph semi-supervised learning, self-supervised learning, and pseudo-labeling to learn prototype representations of new classes in a zero-shot way. In contrast to existing solutions from the vision domain, POWN does not require data augmentation techniques for node classification. Experiments on benchmark datasets demonstrate the effectiveness of POWN, where it outperforms baselines by up to $20\%$ accuracy on the small and up to $30\%$ on the large datasets. Source code is available at https://github.com/Bobowner/POWN.