# On the Prediction Instability of Graph Neural Networks

arXiv: https://arxiv.org/abs/2205.10070 · HTML: https://mlsanity.com/paper/2205.10070/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 24 references that are in the map.
- First posted on arXiv 20 May 2022.

## Builds on (most cited first, 19 of 19)

- [Semi-Supervised Classification with Graph Convolutional Networks](https://mlsanity.com/paper/1609.02907/) (2016)
- [Graph Attention Networks](https://mlsanity.com/paper/1710.10903/) (2017)
- [Fast Graph Representation Learning with PyTorch Geometric](https://mlsanity.com/paper/1903.02428/) (2019)
- [Open Graph Benchmark: Datasets for Machine Learning on Graphs](https://mlsanity.com/paper/2005.00687/) (2020)
- [Similarity of Neural Network Representations Revisited](https://mlsanity.com/paper/1905.00414/) (2019)
- [Pitfalls of Graph Neural Network Evaluation](https://mlsanity.com/paper/1811.05868/) (2018)
- [Revisiting Semi-Supervised Learning with Graph Embeddings](https://mlsanity.com/paper/1603.08861/) (2016)
- [Benchmarking Graph Neural Networks](https://mlsanity.com/paper/2003.00982/) (2020)
- [Adversarial Attacks on Neural Networks for Graph Data](https://mlsanity.com/paper/1805.07984/) (2018)
- [Wiki-CS: A Wikipedia-Based Benchmark for Graph Neural Networks](https://mlsanity.com/paper/2007.02901/) (2020)
- [Certifiable Robustness and Robust Training for Graph Convolutional Networks](https://mlsanity.com/paper/1906.12269/) (2019)
- [Randomness In Neural Network Training: Characterizing The Impact of Tooling](https://mlsanity.com/paper/2106.11872/) (2021)
- [Nondeterminism and Instability in Neural Network Optimization](https://mlsanity.com/paper/2103.04514/) (2021)
- [On the Reproducibility of Neural Network Predictions](https://mlsanity.com/paper/2102.03349/) (2021)
- [Stability of Graph Convolutional Neural Networks to Stochastic Perturbations](https://mlsanity.com/paper/2106.10526/) (2021)
- [Churn Reduction via Distillation](https://mlsanity.com/paper/2106.02654/) (2021)
- [The Effects of Randomness on the Stability of Node Embeddings](https://mlsanity.com/paper/2005.10039/) (2020)
- [Anti-Distillation: Improving reproducibility of deep networks](https://mlsanity.com/paper/2010.09923/) (2020)
- [Model Stability with Continuous Data Updates](https://mlsanity.com/paper/2201.05692/) (2022)

## Built on by (most cited first, 6 of 6)

- [Similarity of Neural Network Models: A Survey of Functional and Representational Measures](https://mlsanity.com/paper/2305.06329/) (2023)
- [Geometric instability of graph neural networks on large graphs](https://mlsanity.com/paper/2308.10099/) (2023)
- [Distilling Influences to Mitigate Prediction Churn in Graph Neural Networks](https://mlsanity.com/paper/2310.00946/) (2023)
- [Exploring End-to-end Differentiable Neural Charged Particle Tracking -- A Loss Landscape Perspective](https://mlsanity.com/paper/2407.13420/) (2024)
- [Revisiting the Relation Between Robustness and Universality](https://mlsanity.com/paper/2510.19427/) (2025)
- [The Impact of Dimensionality on the Stability of Node Embeddings](https://mlsanity.com/paper/2604.08492/) (2026)

## Abstract (arXiv)

Instability of trained models, i.e., the dependence of individual node predictions on random factors, can affect reproducibility, reliability, and trust in machine learning systems. In this paper, we systematically assess the prediction instability of node classification with state-of-the-art Graph Neural Networks (GNNs). With our experiments, we establish that multiple instantiations of popular GNN models trained on the same data with the same model hyperparameters result in almost identical aggregated performance but display substantial disagreement in the predictions for individual nodes. We find that up to one third of the incorrectly classified nodes differ across algorithm runs. We identify correlations between hyperparameters, node properties, and the size of the training set with the stability of predictions. In general, maximizing model performance implicitly also reduces model instability.