The Infinity Mirror Test for Graph Models

TL;DR

The paper introduces the Infinity Mirror test, a method for evaluating the robustness of graph models by repeatedly fitting models to their own outputs, revealing biases and weaknesses.

Contribution

It presents a novel stress test for graph models that exposes implicit biases by iterative self-fitting, aiding in understanding model limitations.

Findings

Conventional graph models often degenerate under the Infinity Mirror test.

The test reveals biases and assumptions in graph models that are not apparent with standard evaluation.

Degenerative patterns observed can guide future development of more robust graph models.

Abstract

Graph models, like other machine learning models, have implicit and explicit biases built-in, which often impact performance in nontrivial ways. The model's faithfulness is often measured by comparing the newly generated graph against the source graph using any number or combination of graph properties. Differences in the size or topology of the generated graph, therefore, indicate a loss in the model. Yet, in many systems, errors encoded in loss functions are subtle and not well understood. In the present work, we introduce the Infinity Mirror test for analyzing the robustness of graph models. This straightforward stress test works by repeatedly fitting a model to its own outputs. A hypothetically perfect graph model would have no deviation from the source graph; however, the model's implicit biases and assumptions are exaggerated by the Infinity Mirror test, exposing potential issues that were previously obscured. Through an analysis of thousands of experiments on synthetic and real-world graphs, we show that several conventional graph models degenerate in exciting and informative ways. We believe that the observed degenerative patterns are clues to the future development of better graph models.