BLOG

On March 15, the second Neurotech Forum took place. The event was organized by IBM and the Institute of Neuroinformatics, ETHZ as part of the Neurotech project, a European Union funded collaborative network of academia, institutes and industry active in neuromorphic computing technology. It attracted 180 participants and generated a lively debate in the online chat.

The aim of the Neurotech forum is to  address some of the most urgent questions in AI.

Recapping the event, the morning began with the presentation of several types of new neuromorphic computing technologies, ranging from novel circuits in today’s silicon CMOS technology to advanced new concepts in electronics and photonics.

An example of the first type was presented by Graphcore, a UK-based company focusing on novel accelerators for neuromorphic computing based on a distributed on-chip memory architecture for enhanced neural network training. The Institute for Neuroinformatics (INI), Synsense and IniVation (Swiss start-ups) then highlighted a range of event-based integrated circuits they’ve been exploring. And smartwatch producer Withings, based in France, presented ways to reduce data processing requirements that would enable local information extraction instead of extracting from the cloud. Along these lines, Italian Institute of Technology (IIT) discussed their efforts on neuromorphic computing technology and algorithms for robotics.

New hardware concepts were presented by LightOn (France), IMEC (Belgium) and IBM. LightOn establishes novel photonic neuromorphic processors for reducing the dimension of the input data for more efficient classification and training. IMEC and IBM presented new analog signal processing hardware concepts for neuromorphic accelerators.

In the afternoon, a discussion took place on ways to enhance the uptake of AI in companies. A presentation by II-VI Laser Enterprise in Zurich, described steps of bringing AI into a laser manufacturing line for process control. The panelists agreed that skepticism and resistance to predefining processes are delaying the application of AI more than the corresponding technical hurdles. They discussed a fast-evolving landscape and the strong need to build a community around neuromorphic computing technology.

The Neurotech Forum closed with a presentation by Marcello Lenca from the ETH Zurich on AI and ethics. Steve Furber led the debate, which touched on  the benefits but also risks of AI.

A big thank-you the co-organisers at INI, Melika Payvand and Elisa Donati (INI) and to Marcel Begert for his excellent technical support.

The IBM organizing team – Laura Bégon-Lours, Antonio La Porta and Bert Offrein.