EDGE AI Neuromorphic WG

Apr 30, 2026

Invited Vijay Janapa Reddi petrut.bogdan@innatera.com jyik@g.harvard.edu steve.furber@manchester.ac.uk eran@weebit-nano.com Pete Bernard jeshragh@ucsc.edu raashid.ansari@silabs.com Rosina Haberl csthakur@iisc.ac.in EAIF Team Calendar benedetto.leto@polito.it massimiliano.versace@analog.com nir.sharvit@analog.com xianju.wang@analog.com gideon@weebit-nano.com klaus.knobloch@infineon.com priya.panda@yale.edu friedemann.zenke@fmi.ch dhireesha.kudithipudi@utsa.edu venkata.rayudu@silabs.com seunghun.shin@murata.com jtapson@brainchip.com frank.ghenassia@ceva-ip.com tinoosh@jhu.edu daniel.cavalcanti@analog.com rahim@dsv.su.se Ed Doran r.jolivet@maastrichtuniversity.nl andre.vanschaik@manchester.ac.uk

Summary

The team aligned on external neuromorphic organization partnerships and finalized the fall live stream scheduling strategy.

Neuromorphic External Partnerships
Collaborations with Project Phaser and Open Neuromorphics will focus on providing industrial steering and guidance. The working group will act as a central clearing house for neuromorphic projects.

Scheduling Fall Live Stream
The fall live stream is tentatively set for September 16. Discussions included adding mobility and automotive categories to the upcoming call for presentations.

Content Strategy and Updates
Generative artificial intelligence will convert live stream content into multimodal formats. A large scale project update was shared regarding the Spinnaker 2 hardware implementation at Jülich.

Decisions

  • Industrial support for neuromorphic organizations The working group will provide industrial steering and guidance to the Project Phaser and Open Neuromorphics organizations.
  • Design contest for team merchandise The group will conduct a rock-themed design contest for the creation of group t-shirts and stickers.
  • September live stream date scheduling The next live stream is scheduled for Wednesday, September 16th.
  • Dedicated community sessions for live stream The group will host dedicated, invitation-only sessions for neuromorphic communities (specifically Project Phaser and Open Neuromorphics) during the live stream, rather than accepting general submissions for these topics.

Next steps

  • unticked[Rosina Haberl] Share Collaboration Info: Send information packets and university promotion codes to Steve. Email Steve regarding potential UK/London engagement.
  • unticked[Pete Bernard] Organize Design Contest: Create a design contest for rock-themed t-shirts/stickers.
  • unticked[Rosina Haberl] Distribute CFP: Finalize abstract requirements in the document. Send out the Call for Presentations next week.
  • unticked[Pete Bernard] Secure Mercedes Contact: Continue efforts to find a Mercedes contact. Discuss neuromorphic AI usage.

Details

  • Initial Check-in and Coordination for UK Engagement: Rosina Haberl initiated the meeting with a casual check-in and discussed Steve Furber’s location, confirming they are in the UK, about 100 miles from London, near Derby. Rosina Haberl suggested Steve Furber could join an event in London and offered to share community codes for university students they know, specifically mentioning connections in Manchester. Steve Furber accepted the offer of information but noted that travel is getting difficult due to needing medical oxygen for 15 hours a day.
  • Call for Papers/Presentations and Attendance Issues: The group noted the low attendance for the current call, and Rosina Haberl proposed moving forward with the call for papers regardless of the absence of others. Petrut Bogdan indicated they cannot force attendees to work, as they are not their employees. Pete Bernard arrived, and Rosina Haberl informed them that Steve Furber had agreed to help disseminate the word about the call within the UK and at universities.
  • Discussion of External Neuromorphic Organizations: Petrut Bogdan brought up an email they sent regarding external organizations, specifically Project Phaser and Open Neuromorphics, noting that these groups are working on neuromorphic datasets and organization. Project Phaser is creating a tool to look at curated neuromorphic datasets, and Open Neuromorphics is managed by an executive committee (EC) of three researchers who are attempting to establish a foundation. Petrut Bogdan suggested that the AJ Foundation might be able to assist these organizations.
  • Clarification on Inter-Organizational Leadership: Pete Bernard asked if Project Phaser is part of Open Neuromorphics, and Petrut Bogdan explained the connection using an analogy involving the historical unification of Romanian regions. Petrut Bogdan clarified that a shared leader, referred to as “Blessing,” is involved with both organizations, making them a common denominator and a good contact point for the group. The group discussed Project Phaser’s output, which includes a platform for curated datasets, and their other goals related to common compilers for neuromorphic backends and creating a platform for uploading simulators.
  • Identifying Needs of External Organizations: Petrut Bogdan noted that the people uploading datasets to Project Phaser’s platform, called “episodes,” need guidance on hardware implications, as they focus on the efficiency of spike-coded data without considering the complexity and energy wasted in the encoding conversion from real data. The hardware-specific aspects of implementation are not entirely clear to them. Pete Bernard also noted that Open Neuromorphics has a directory of various projects and that the challenge lies in providing organizational energy and focus to the existing work.
  • Potential Role of the Working Group with External Organizations: The group discussed how their neuromorphic working group could assist these organizations by providing industrial steering and guidance to ensure projects are more commercially useful and potentially secure funding from commercial entities. Pete Bernard suggested the working group could serve as a “central clearing house” for various neuromorphic projects, helping to break down barriers between communities. Pete Bernard mentioned that they had suggested a session with Blessing to see how the pieces could fit together.
  • Scheduling the Fall Live Stream and Call for Presentations: The conversation shifted to scheduling the fall live stream, with Petrut Bogdan proposing September 15th, a Tuesday, which is before the Things Conference in Amsterdam on September 22nd–23rd. Rosina Haberl suggested that Wednesdays generally work well for live streams, and the group tentatively agreed to Wednesday, September 16th. The group also discussed the possibility of doing a precursor event before the Things Conference and having a session at the conference itself.
  • Developing Multimodal Content Strategy: Petrut Bogdan and Pete Bernard discussed creating a perspective paper and other multimodal content (like a podcast or paper derived from the live stream) to reach a wider audience. Pete Bernard proposed using generative AI (Gen AI) to help convert the content into various formats, while Petrut Bogdan humorously suggested creating “boring” content channels for sleep.
  • Refining Live Stream Content Categories: The group reviewed the categories for the call for presentations, aiming for a more categorized approach based on use cases. Pete Bernard proposed adding “Mobility and Automotive” as a specific topic after reaching out to Mercedes to understand their application of neuromorphic AI. Rosina Haberl will get the call for papers document finalized and sent out next week.
  • Strategy for Community and Tooling Sessions: The group confirmed that the live stream should include dedicated sessions on communities and tooling, specifically mentioning Project Phaser, Open Neuromorphics, and the Tour Consortium. Petrut Bogdan emphasized that these sessions should feature specific, invited groups rather than being open to random submissions due to the limited number of established groups in this area.
  • Update on Jülich (Jülich Supercomputing Centre) Spinnaker System: Steve Furber shared an update on a launch event for a large project at the University of ULI in Germany, which is utilizing two racks of Spinnaker 2 hardware. The system is being applied to event-based applications, finite element modeling, and computation, in addition to neuroscience. Guido Trench is leading the project bid at Jülich.

How is the quality of these specific notes? Take a short survey to let us know your feedback, including how helpful the notes were for your needs.