Focus Issue on Statistical Inference and Machine Learning for Complex Networks

Guest Editors

• George Cantwell, University of Cambridge, United Kingdom
• Caterina De Bacco, Max Planck Institute for Intelligent Systems, Germany
• Jean-Gabriel Young, University of Vermont, USA
• Alec Kirkley, University of Hong Kong, Hong Kong

Scope

Complex networks present a wide range of challenges for statistical inference and machine learning due to their unique structural and dynamical properties. Networks are relational, discrete, strongly correlated, and high-dimensional by nature, making their structure not easily amenable to traditional inference and learning methods that are tailored for continuous, ordered, and/or weakly interacting systems. The dynamics on and of networks faces the same set of challenges, with additional issues related to creating meaningful temporal representations of the dynamics that retain meaningful system structure while permitting analysis with the rich set of tools available for sparse network data. Moreover, real complex networks often demonstrate unique physical characteristics including clustering, assortativity, transitivity, short paths, and bursty interaction patterns among other features associated with complex systems such as scale-free structure, emergent order, and chaotic dynamics. It is therefore of great importance that we develop statistical models and measures aimed at capturing the intricate structure and dynamics of complex network data in various disciplines, from epidemiology to sociology to urban planning among many other areas.

In order to effectively model networks in real-world applications, it is critical that we develop sound methodology for performing inference, prediction, and evaluation of network models. This often requires new inference and learning objectives as well as optimization and sampling techniques to highlight the patterns of interest in the presence of a high level of heterogeneity and noise. Existing methods for complex network inference and machine learning often suffer from issues with scalability, interpretability, and robustness under different learning environments, so further research refining these methods is also of high importance.

This Focus Issue is concerned with all aspects of statistical inference and machine learning for complex network data and models. Topics to be addressed can include, but are not limited to:

• Theory and algorithms for network inference, including community detection, network ranking, network regression, link prediction, representation learning, and other supervised or unsupervised tasks
• Methods for improving the evaluation and interpretation of network models or inference methods, including network visualization, comparison, or summarization methods
• Generative models for network structure or dynamics
• Studies of algorithm performance and limitations in the context of network inference
• Models and inference/learning methods for higher order structure and dynamics, including hypergraphs and simplicial complexes

Submission process

We encourage submissions from all authors whose work fits with the scope of this focus collection. The collection will also feature invited contributions. All focus issue articles are subject to the same review process as regular articles. Authors are invited to contact one of the guest editors, or the journal team directly, to discuss the suitability of their work prior to submission.

Please submit your article via our online submission form. In Step 1, where the form asks for the article type please select the appropriate option. At the bottom of the page please then select “Focus Issue on Statistical Inference and Machine Learning for Complex Networks” in the ‘Focus Issue’ drop down box.

Deadline for submissions

The target deadline for submissions is 31 May 2025 though we can be flexible where necessary. We encourage early submission where possible, as articles will be published on acceptance without being delayed by other papers in the collection.

Peer review

All focus issue articles will be peer reviewed in the same manner and to the same high standard as regular issue articles, with the peer review overseen and administered by our in-house journal editorial team. Find out more about peer review at IOP Publishing.

Source and more details: https://iopscience.iop.org/collections/jpcomplex-240806-630