25 November 2025 | 11:00 -12.30 CET

Register here: https://us02web.zoom.us/meeting/register/x4NBf1txTbiXBEe-PVqwmA

As data-driven decision-making becomes ever more critical across sectors such as healthcare and finance, AI systems must be able to anticipate complex, fast-changing events with speed, accuracy, and transparency.

The EVENFLOW project is advancing this frontier by developing hybrid learning methods that combine deep learning with logic-based reasoning, resulting in more scalable, explainable, and trustworthy AI.

Join us for the EVENFLOW Webinar to explore the open-source tools created within the project, including:

• EVENFLOW Scalability Toolkit will be demonstrated by N. Giatrakos, ARC
• Reasoning-Based Forecast Interpreter will be demonstrated by N. Katzouris, NCSRD
• EVENFLOW Verification Tool will be demonstrated by S. Varghese & A. Hekal, ICL

EVENFLOW Scalability Toolkit: In this demo it will be presented SuBiTO. SuBiTo is a framework for continuous real-time neural learning over big streaming data that uses synopsis structures to balance training time and accuracy on the fly. It monitors the data stream, detects concept drifts, and interactively recommends optimal NN architectures, synopsis sizes, and epochs through a live dashboard for human-in-the-loop deployment.

Neurosymbolic CER optimised over IoT platforms: In this demo it will be presented NeuroFlinkCEP. NeuroFlinkCEP is a neurosymbolic framework that combines neural models with FlinkCEP’s symbolic complex event recognition to process streaming data at scale. It translates extended-regular-expression patterns into distributed FlinkCEP jobs per IoT device and uses an IoT optimiser to place operators efficiently across edge and cloud.

Reasoning-Based Forecast Interpeter: It will be demonstrated methods developed in the context of the EVENFLOW project for (a) training perception neural networks alongside temporal knowledge for improved generalisation and interpretability; (b)  taking advantage of the hybrid models’ formal probabilistic semantics  to derive human-understandable explanations of the models’ predictions. 

EVENFLOW Verification Toolkit: This demo presents the Verification toolkit developed for verifying Neuro-Symbolic systems, providing guarantees on the model robustness against adversarial perturbations. We will demonstrate how the tool is being used to verify models developed for the Industry 4.0 use case of automated guided vehicle (AGV) navigation and planning.

These innovations support data scientists, AI engineers, and organisations seeking to build more reliable and transparent AI systems.

Register now to discover how EVENFLOW is shaping the future of complex event forecasting: https://us02web.zoom.us/meeting/register/x4NBf1txTbiXBEe-PVqwmA