# Inductive Graph Layout with Implicit Neural Fields

arXiv: https://arxiv.org/abs/2608.08876 · HTML: https://mlsanity.com/paper/2608.08876/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 30 references that are in the map.
- First posted on arXiv 9 Aug 2026.

## Builds on (most cited first, 9 of 9)

- [FiLM: Visual Reasoning with a General Conditioning Layer](https://mlsanity.com/paper/1709.07871/) (2017)
- [Learning Continuous Image Representation with Local Implicit Image Function](https://mlsanity.com/paper/2012.09161/) (2020)
- [Visualizing Large-scale and High-dimensional Data](https://mlsanity.com/paper/1602.00370/) (2016)
- [DeepGD: A Deep Learning Framework for Graph Drawing Using GNN](https://mlsanity.com/paper/2106.15347/) (2021)
- [A Sparse Stress Model](https://mlsanity.com/paper/1608.08909/) (2016)
- [DRGraph: An Efficient Graph Layout Algorithm for Large-scale Graphs by Dimensionality Reduction](https://mlsanity.com/paper/2008.07799/) (2020)
- ["Normalized Stress" is Not Normalized: How to Interpret Stress Correctly](https://mlsanity.com/paper/2408.07724/) (2024)
- [CoRe-GD: A Hierarchical Framework for Scalable Graph Visualization with GNNs](https://mlsanity.com/paper/2402.06706/) (2024)
- [Word2VecGD: Neural Graph Drawing with Cosine-Stress Optimization](https://mlsanity.com/paper/2509.17333/) (2025)

## Abstract (arXiv)

A graph layout is normally a table of N free coordinates. We optimise a function with a fixed number of parameters instead. This gives a drawing a sample complexity and an extensible domain. Force-directed algorithms remain the standard tools for graph drawing. The most accurate among them minimise stress in the Kamada-Kawai formulation by directly optimising the node coordinates, at a full objective cost of O(N2) in time and space. Here, we propose Fling (Field Layout via Implicit Neural Geometry), a small neural network mapping the distances of each node to a set of landmarks, positioning it in the plane by training on the layout energy. The full spring system then becomes tractable without its distance matrix, as rest lengths follow from a landmark bound in constant time per pair while a second network learns the majorisation sums from exact anchor rows, at O(|A|N) per step for |A|ll N anchors. Unlike neural drawers that read the graph by message passing, we represent the drawing as a function of node features. An unseen node costs one forward pass, where sparse and low-rank majorisation remain transductive. As the unknowns are weights rather than coordinates, the energy only requires a small fraction of the nodes, and a field fitted that way outperforms PivotMDS, landmark MDS, and a kernel ridge trained on the same energy and features, when the task is fitting the energy of a graph from a sample of its nodes. In addition, the same parameterisation enables a stochastic pivot stress variant, an aesthetics-optimised variant carrying a neighbour-embedding energy with node-edge clearance and crossing terms on the same field, and conditioning on the weight between two energies gives a whole layout family from one run.