Recurrent Attention Walk for Semi-supervised Classification
Uchenna Akujuobi, Qiannan Zhang, Han Yufei, Xiangliang Zhang
TL;DR
We address semi-supervised node classification on attributed graphs by learning a targeted neighborhood exploration strategy using Recurrent Attention Walk (RAW). RAW casts neighborhood exploration as a finite-horizon POMDP and optimizes a policy via reinforcement learning, integrating a score network, a GRU-based core, and a classifier to produce a trajectory embedding $h_T$ for the starting node. The method operates in both transductive and inductive settings, reduces memory and noise by selectively walking rather than aggregating all neighbors, and provides interpretable walk trajectories. Empirical results on four citation datasets show RAW achieving competitive or superior accuracy compared to state-of-the-art baselines, with favorable scalability and meaningful trajectory analyses.
Abstract
In this paper, we study the graph-based semi-supervised learning for classifying nodes in attributed networks, where the nodes and edges possess content information. Recent approaches like graph convolution networks and attention mechanisms have been proposed to ensemble the first-order neighbors and incorporate the relevant neighbors. However, it is costly (especially in memory) to consider all neighbors without a prior differentiation. We propose to explore the neighborhood in a reinforcement learning setting and find a walk path well-tuned for classifying the unlabelled target nodes. We let an agent (of node classification task) walk over the graph and decide where to direct to maximize classification accuracy. We define the graph walk as a partially observable Markov decision process (POMDP). The proposed method is flexible for working in both transductive and inductive setting. Extensive experiments on four datasets demonstrate that our proposed method outperforms several state-of-the-art methods. Several case studies also illustrate the meaningful movement trajectory made by the agent.