Lifelong Graph Learning
This addresses the need for lifelong learning in graph neural networks for domains like wearable devices and feature matching, though it is incremental by adapting existing lifelong learning techniques from CNNs to GNNs.
The paper tackles the problem of learning graph-structured data continuously in a streaming fashion by proposing a feature graph network (FGN) that converts continual graph learning into a regular graph learning problem, achieving superior performance in applications like lifelong human action recognition and feature matching.
Graph neural networks (GNN) are powerful models for many graph-structured tasks. Existing models often assume that the complete structure of the graph is available during training. In practice, however, graph-structured data is usually formed in a streaming fashion so that learning a graph continuously is often necessary. In this paper, we bridge GNN and lifelong learning by converting a continual graph learning problem to a regular graph learning problem so GNN can inherit the lifelong learning techniques developed for convolutional neural networks (CNN). We propose a new topology, the feature graph, which takes features as new nodes and turns nodes into independent graphs. This successfully converts the original problem of node classification to graph classification. In the experiments, we demonstrate the efficiency and effectiveness of feature graph networks (FGN) by continuously learning a sequence of classical graph datasets. We also show that FGN achieves superior performance in two applications, i.e., lifelong human action recognition with wearable devices and feature matching. To the best of our knowledge, FGN is the first method to bridge graph learning and lifelong learning via a novel graph topology. Source code is available at https://github.com/wang-chen/LGL