Stochastic Gradient Descent Works Really Well for Stress Minimization

TL;DR

This paper evaluates a stochastic gradient descent approach for stress minimization in graph layouts, finding it simpler and more robust despite not producing better layouts than existing methods.

Contribution

It provides experimental evidence that stochastic gradient descent is a viable alternative to majorization, emphasizing its simplicity and robustness over layout quality.

Findings

Stochastic gradient descent does not outperform majorization in layout quality.

The new approach is simpler to implement and more robust to poor initializations.

Experimental results favor the stochastic gradient descent method for practical use.

Abstract

Stress minimization is among the best studied force-directed graph layout methods because it reliably yields high-quality layouts. It thus comes as a surprise that a novel approach based on stochastic gradient descent (Zheng, Pawar and Goodman, TVCG 2019) is claimed to improve on state-of-the-art approaches based on majorization. We present experimental evidence that the new approach does not actually yield better layouts, but that it is still to be preferred because it is simpler and robust against poor initialization.