The second AIDA4Edge Training School, entitled “Reliable and Efficient AI Hardware Accelerators”, was successfully organized from May 11–13, 2026, at IHP – Leibniz Institute for High Performance Microelectronics, Frankfurt (Oder), Germany, within the Horizon Europe Twinning project AIDA4Edge.
The event opened with the Opening Session and IHP Welcome Address, delivered by Marko Andjelković and Fabian Vargas, followed by an introduction to the AIDA4Edge project presented by Milan Dinčić, AIDA4Edge project coordinator from the Faculty of Electronic Engineering, University of Niš.
The event gathered researchers, professors, PhD students, and industry experts working in the fields of Edge AI, hardware accelerators, neuromorphic computing, reliability-aware design, and open-source EDA methodologies. The training school provided participants with an opportunity to gain insights into the latest advances in reliable and energy-efficient AI hardware systems through keynote talks, technical lectures, and interactive discussions. The event featured a rich three-day program of keynote lectures, technical presentations, project meetings, and networking activities. The detailed program is available here (LINK).
The training school featured keynote talks by distinguished international experts. Artur Jutman (Tallinn University of Technology, Estonia) opened the scientific program with a lecture on reusing Design-for-Test hardware for self-health awareness and fault recovery in multicore CPU subsystems, highlighting new approaches to runtime resilience in complex computing systems. Mahdi Taheri (BTU Cottbus–Senftenberg, Germany) presented runtime-adaptive cross-level optimization techniques for efficient and reliable Edge AI, emphasizing coordinated adaptivity across model, hardware, and arithmetic levels.
The program included several lectures delivered by researchers from IHP, covering advanced AI hardware technologies and reliability-aware design methodologies. Emilio Perez-Bosch Quesada, Jianan Wen, and Andre Lucas Chinazzo presented research on RRAM technologies, in-memory computing architectures, frequency-domain CNN accelerators, and fault propagation in computing-in-memory systems, highlighting challenges and opportunities in reliable AI acceleration.
A strong emphasis was placed on brain-inspired and neuromorphic hardware systems. Leticia Bolzani Poehls (IHP) discussed hybrid brain-inspired architectures and technology heterogeneity for high-performance AI applications, while Davide Bertozzi (University of Manchester, UK) delivered a keynote lecture addressing the question “When Hardware Fails, Does Intelligence Survive?”, focusing on faults in neuromorphic communication hardware and cross-layer reliability analysis in neuromorphic systems.
Additional highlights included the presentation by Christos Sotiriou, our stakeholder from the TWIN-RELECT project (University of Thessaly, Greece), who shared recent advances and future perspectives in EDA Agentic AI; Krzysztof Herman (IHP), discussing developments in open-source EDA and PDK ecosystems; and Rizwan Tariq Syed (IHP), reviewing fault injection methodologies for neural network accelerators.
Beyond technical advances, the school also addressed broader aspects of research and innovation. Fabian Vargas (IHP) presented approaches for mixed-criticality task execution in multicore real-time systems, while Anders Henriksson (IHP Solutions) and Milos Krstić (IHP) shared valuable insights on commercialization of academic research results and preparation of international project proposals.
The training school additionally served as an important platform for strengthening collaboration among AIDA4Edge partners: Faculty of Electronic Engineering, University of Niš, IHP – Leibniz Institute for High Performance Microelectronics, University of Manchester, and University of Ferrara. Project meetings held during the event further supported scientific exchange, knowledge transfer, and planning of future collaborative activities. The event was organized within the AIDA4Edge project, funded by the European Union’s Horizon Europe research and innovation programme under Grant Agreement No. 101160293, with additional support from project partners and sponsors.
