Deep Learning for Graphs is a special session at the 2022 International Joint Conference on Neural Networks, World Congress on Computational Intelligence that will be held in Padua, Italy on July 18-23, 2022.

Call for papers

Graph data processing and deep learning are two essential subfields of machine learning research. Each of them has provided unprecedented advancements in different research fields and exciting breakthroughs, from the analysis of scientific experimental data to industrial applications in computer vision and recommendation systems.

Graphs are a general and complex way of representing structured data, allowing us to effectively describe systems of interacting elements, like social, biological, and technological networks, as well as data where topological variations influence the feature of interest, e.g., the interaction of proteins or molecular compounds. Graphs also provide a means to insert prior knowledge and inductive biases, such as symmetries and constraints, into the data representation to reduce the parameter search space or to improve the learning process.

Representing data in this rich structured form provides a fundamental advantage for identifying patterns suitable for predictive or exploratory data analysis. This has motivated a recent increasing interest of the machine learning community in the development of learning models for structured information.

The field of graph deep learning, in particular, combines the ability of deep neural networks to learn representations end-to-end with this explicit description of relations in the data. Specifically, the class of models at the heart of graph deep learning, generically called Graph Neural Networks (GNNs), extend and generalize typical convolutional neural networks to process arbitrary graphs.

Topics

This session focuses on the broad spectrum of machine learning methods for structured and relational data, with a focus on deep representation learning. Topics of interest to this session include, but are not limited to:

• Graph representation learning

• Graph generation (probabilistic models, variational autoencoders, adversarial learning, etc.)

• Graph learning and relational inference

• Graph coarsening and pooling in graph neural networks

• Graph kernels and distances

• Scalability, data efficiency, and training techniques of graph neural networks

• Tensor methods for structured data

• Theory of graph neural networks (e.g., expressive power, learnability, negative results)

• Relational reinforcement learning

• Deep learning for dynamic graphs and graph sequences

• Anomaly and change detection in graph data

• Reservoir computing and randomized neural networks for graphs

• Recurrent, recursive and contextual models

• Graph datasets and benchmarks

• Applications in natural language processing, computer vision (e.g. point clouds), materials science, cheminformatics, computational biology, social networks, etc.

Important Dates

Paper Submissions: January 31, 2022 (11:59 PM AoE) Updated!
Title and Abstract Submission: January 31, 2022 (11:59 PM AoE). (New submissions cannot be created past this deadline.)
Complete Paper (pdf) Submission: February 14, 2022 (11:59 PM AoE)

Paper Acceptance Notifications: April 26, 2022

Conference: July 18-23, 2022

Session Organisers

• Davide Bacciu, University of Pisa (IT)

• Shirui Pan, Monash University (AU)

• Daniele Grattarola, IDSIA, Università della Svizzera italiana (CH)

• Miao Zhang, Aalborg University (DK)

• Nicolò Navarin, University of Padova (IT)

• Feng Xia, Federation University Australia (AU)

• Daniele Zambon, IDSIA, Università della Svizzera italiana (CH)

For any enquire, please write to daniele.zambon [at] usi.ch or daniele.grattarola [at] usi.ch .