Hierarchical Prototype Networks for Continual Graph Representation Learning

TL;DR

This paper introduces Hierarchical Prototype Networks (HPNs) for continual graph learning, enabling models to adapt to new node categories while preventing forgetting of previous data through a hierarchical prototype approach.

Contribution

The paper proposes a novel hierarchical prototype framework with bounded memory and theoretical guarantees to prevent catastrophic forgetting in continual graph learning.

Findings

HPNs outperform state-of-the-art methods on five datasets.

Memory consumption of HPNs remains bounded regardless of task number.

Theoretical proofs show no forgetting of previous prototypes during learning new tasks.

Abstract

Despite significant advances in graph representation learning, little attention has been paid to the more practical continual learning scenario in which new categories of nodes (e.g., new research areas in citation networks, or new types of products in co-purchasing networks) and their associated edges are continuously emerging, causing catastrophic forgetting on previous categories. Existing methods either ignore the rich topological information or sacrifice plasticity for stability. To this end, we present Hierarchical Prototype Networks (HPNs) which extract different levels of abstract knowledge in the form of prototypes to represent the continuously expanded graphs. Specifically, we first leverage a set of Atomic Feature Extractors (AFEs) to encode both the elemental attribute information and the topological structure of the target node. Next, we develop HPNs to adaptively select relevant AFEs and represent each node with three levels of prototypes. In this way, whenever a new category of nodes is given, only the relevant AFEs and prototypes at each level will be activated and refined, while others remain uninterrupted to maintain the performance over existing nodes. Theoretically, we first demonstrate that the memory consumption of HPNs is bounded regardless of how many tasks are encountered. Then, we prove that under mild constraints, learning new tasks will not alter the prototypes matched to previous data, thereby eliminating the forgetting problem. The theoretical results are supported by experiments on five datasets, showing that HPNs not only outperform state-of-the-art baseline techniques but also consume relatively less memory.