Seminar | Self-Designing Edge Intelligence for Physical AI – 30 November 2026

Dove e quando

📅 Date: Monday, November 30, 2026
🕒 Time: 4:00 PM (CET)
📍 Venue: Room D3, DITEN – University of Genoa, Via Opera Pia 11a, 16145 Genova
💻 Online: join on Microsoft Teams
📝 Registration: attendance is free and no registration is required

The IEEE Circuits and Systems Society (CASS) North Italy Chapter and DITEN – University of Genoa are pleased to invite PhD students and researchers to the seminar Self-Designing Edge Intelligence for Physical AI, given by Andrea Mattia Garavagno (ETH Zürich, Switzerland).

Programme

  • 16:00–16:05 – Welcome and introduction, Prof. Edoardo Ragusa, University of Genoa (Session Chair)
  • 16:05 – Seminar by Andrea Mattia Garavagno, ETH Zürich, Switzerland

Abstract

In Physical AI systems, comprehensive training data are rarely available prior to deployment, as collecting rare physical events, fault states, or operational anomalies in advance is often prohibitively expensive or impractical. Consequently, deployed edge models must evolve alongside the gradual, operational accumulation of sensor data. However, conventional edge intelligence remains static, requiring substantial manual intervention and cloud resources to incorporate newly observed data.

This seminar presents a framework for self-designing edge intelligence that distributes neural design directly across the system lifecycle. Leveraging human-in-the-loop validation and lightweight, derivative-free hardware-aware neural architecture search (HW-NAS), the system autonomously adapts and redesigns deep learning models locally on embedded platforms. Operating entirely at the edge without cloud dependency, this paradigm ensures data sovereignty and resource efficiency for resilient cyber-physical systems and intelligent sensor interfaces.

Speaker

Andrea Mattia Garavagno is a Postdoctoral Researcher at ETH Zurich, Switzerland. He earned his Ph.D. from the University of Genoa (co-funded by Scuola Superiore Sant'Anna) and served as a Harvard SEAS Fellow in 2025. He co-authored the book "Introduction to Microprocessor-Based Systems Design" (Springer, 2022). His research bridges computer architecture, machine learning, and embedded systems for self-designing edge intelligence.

Organisers

Professors Edoardo Ragusa, Paolo Gastaldo and Maurizio Valle, IEEE CASS North Italy Chapter.

The seminar flyer and programme are available in the attachments below.

Contatti

Session Chair: Prof. Edoardo Ragusa, University of Genoa – edoardo.ragusa@unige.it

Last update