A two-day workshop in Neural Network Safety at The National Robotarium, Heriot-Watt University. Part of our Robotics CPD programme.

Course overview

Dates: 22-23 March 2027

Venue: The National Robotarium, Heriot-Watt University, Edinburgh

Format: Two-day in-person tutorial with lectures and practical workshops

Audience: Industry professionals, researchers, academics and postgraduate students working in AI, robotics and safety-critical systems.

Delivered by Professor Ekaterina Komendantskaya, Alistair Sirman and contributors including Dr Matthew Daggitt, the course uses the open-source Vehicle specification language to bridge the gap between human-defined requirements and automated verification tools.

A practical, hands-on learning experience

Approximately half of the course is dedicated to hands-on exercises and guided labs.

Participants will work with real verification tools and machine learning models, gaining experience in:

Writing verification specifications

Verifying neural networks against safety requirements

Training image classifiers using formal specifications

Evaluating robustness properties

Applying verification workflows to real-world AI systems

All course materials, including slides, notebooks, datasets and trained models, are provided. Exercises can be completed using Google Colab or a local Python installation.

Lunch and refreshments are included throughout the event.

Visit our website (https://thenationalrobotarium.com/industry/robotics-training-development/robotics-cpd/robotics-cpd-neural-network/) for more information.

Description supplied by the event organiser.