During our visit to the University of Manchester, we attended the Training in Reliable Electronic Design and Reliable Neuromorphic Computing, jointly delivered by the TWIN-RELECT and AIDA4Edge projects. The training was organized in close collaboration with our stakeholder, the TWIN-RELECT team, which is also one of the partners in the proposal submitted to the European Electronics Initiative (EEI) call, with FEEUNI participating as well.

The program featured a series of keynote and invited talks, including neutron beam irradiation for electronic testing by Carlo Cazzaniga (STFC Rutherford Appleton Laboratory), design and validation of chips for space applications by Miloš Krstić (IHP Microelectronics), and bit-precise neural network verification by Edoardo Manino (University of Manchester). Reliability of AI accelerators was addressed by experts from TalTech and IHP Microelectronics, while Alessandro Veronesi (IHP, in collaboration with UOM) presented cycle-accurate HW/SW co-design methodologies and led hands-on sessions on fault simulation using the EMBER framework.

In the area of neuromorphic computing, the training included tutorials on Spiking Neural Networks tools by Oliver Rhodes, neuromorphic hardware platforms by Davide Bertozzi (UOM), a visit to the SpiNNaker system hosted by Andrew Rowley, and a keynote on safe and reliable neuromorphic computing by Haralampos Stratigopoulos (Sorbonne Université, CNRS, LIP6). Additional sessions covered reliable AER routing, asynchronous NoCs, SNN training with snnTorch, and SNN–hardware fault co-simulation. Overall, the training provided an excellent balance of advanced technical content and practical hands-on experience, while fostering deeper connections with colleagues from the TWIN-RELECT project through a better understanding of our joint and complementary research interests.