1: \begin{abstract}

2: Popular graph neural networks are shallow models, despite the success of very deep architectures in other application domains of deep learning.

3: This reduces the modeling capacity and leaves models unable to capture long-range relationships.

4: The primary reason for the shallow design results from over-smoothing, which leads node states to become more similar with increased depth.

5: We build on the close connection between GNNs and PageRank, for which personalized PageRank introduces the consideration of a personalization vector.

6: Adopting this idea, we propose the Personalized PageRank Graph Neural Network (PPRGNN), which extends the graph convolutional network to an infinite-depth model that has a chance to reset the neighbor aggregation back to the initial state in each iteration.

7: We introduce a nicely interpretable tweak to the chance of resetting and prove the convergence of our approach to a unique solution without placing any constraints, even when taking infinitely many neighbor aggregations. As in personalized PageRank, our result does not suffer from over-smoothing.

8: While doing so, time complexity remains linear while we keep memory complexity constant, independently of the depth of the network, making it scale well to large graphs. 

9: We empirically show the effectiveness of our approach for various node and graph classification tasks. PPRGNN outperforms comparable methods in almost all cases.

10: \footnote{Our code is available at: https://github.com/roth-andreas/pprgnn}

11: \end{abstract}

12: