WORKSHOP
“Graph Neural Networks for Molecules and Materials”
We bring together leading academic scientists and researchers!
October 2nd, 2026 1pm - 5pm
Scope of the event
Molecules and crystals are naturally graphs: atoms are nodes, and bonds or spatial proximity define edges. This structural correspondence makes graph neural networks the architecture of choice for learning structure–property relationships across chemistry and materials science. This workshop traces the development of GNNs from the foundational Message Passing Neural Network framework through to crystal graph convolutions, directed message passing, and modern equivariant architectures that encode rotational and translational symmetry directly into the network.
The session examines Crystal Graph Convolutional Neural Networks (CGCNN) and MEGNet for solid-state property prediction, directed message passing networks (Chemprop) for molecular property and reaction predic- tion, and the extension to equivariant GNNs that respect the geometric symmetries of three-dimensional atomic arrangements. Applications span molecular property prediction, crystal stability and band gap estimation, re- action outcome forecasting, and materials design. Benchmark datasets — QM9 for small organic molecules, the Materials Project for inorganic crystals, and emerging electrochemical datasets — are discussed both as practical resources and as illustrations of the strengths and limitations of current evaluation methodology.
Participants will gain a comprehensive understanding of how graph-based architectures encode chemical struc- ture, the role of symmetry in determining model accuracy and data efficiency, and the current capabilities and limitations of GNN-based property prediction.
Extended Key Takeaways & Outputs
• Understanding of the Message Passing framework and its extensions for molecules and crystals
• Comparative assessment of CGCNN, MEGNet, Chemprop, and equivariant GNN architectures
• Practical knowledge of benchmark datasets and their strengths and limitations
• Insight into the role of rotational and translational equivariance in model performance
• Foundation for the interatomic potential methods addressed in Workshop 6
Agenda & Speakers
13:00–13:05 — Introduction & Objectives
Organizer
· Why molecules and materials are naturally represented as graphs
· From traditional descriptors to graph-based learning
· Workshop objectives: structure-aware modelling and symmetry-informed learning
13:05–13:35 — Message Passing Neural Networks: Foundations and Intuition
Speaker (Academia): N.N.
· Nodes, edges, and message passing in graph representations
· Learning structure–property relationships from atomic connectivity
13:35–14:05 — Graph Neural Networks for Materials: CGCNN and MEGNet
Speaker (Academia): N.N.
· Crystal Graph Convolutional Neural Networks (CGCNN)
· MEGNet framework and applications to solid-state materials
14:05–14:35 — Molecular Graph Models and Directed Message Passing
Speaker (Academia): N.N.
· Directed message passing neural networks (Chemprop)
· Applications in molecular property prediction and reaction modelling
14:35–14:45 — Break
14:45–15:15 — Equivariant Graph Neural Networks and Geometric Deep Learning
Speaker (Industry): N.N.
· Encoding rotational and translational symmetry in GNNs
· Advantages for 3D molecular and materials systems
15:15–15:45 — Applications and Benchmark Datasets in Chemistry and Materials Science
Speaker (Industry): N.N.
· QM9, Materials Project, and emerging datasets
· Strengths and limitations of current benchmarking practices
15:45–16:15 — Data Efficiency, Generalisation, and Model Limitations
Speaker (Policy/Applied Research): N.N.
· Data requirements and transferability across chemical space
· Limitations of current GNN approaches in real-world applications
16:15–16:55 — Panel Discussion & Strategic Alignment
All Speakers + Moderator
· Are GNNs the dominant paradigm for chemistry and materials ML?
· Balancing accuracy, interpretability, and computational cost
· Role of symmetry and physics-informed constraints
· Future directions: integration with simulations and experiments
Concluding remarks
Registration
Deadlines:
Early-Bird: until July 15th, 2026
Registration : until September 5th, 2026
The registration takes place online by filling out the form.
The registration and access to the virtual room for workshop is effective only if the fees payment has been confirmed.
Please note that INVIRTA MEMBERS can purchase all our workshops and services with 10% discount . If you desire to get more information on our membership please press here .
Prices* :
Industry: 150 € (early bird**: 125 €)
Academy: 125 € (early bird**: 100 €)
Students***: 90 € (early bird**: 75 €)
* Note that prices are subject to VAT (for European countries, except Germany, provide the VAT number of your organisation to be within the VAT reverse framework payment).
** Until November 15th, 2020
*** Copy of the registration certificate (Ph.D, undergraduates)
Special price offers for Workshop packages
Our Workshop packages are valid for any workshops offered by our academy and can be used anytime. If you want to use one of our packages press here.
