AICAS is the IEEE International Conference on Artificial Intelligence Circuits and Systems. It was held for the first time this year, more precisely on March 18-20, in Hsinchu, Taiwan. AICAS' goal is to facilitate research, innovation and development of circuits and systems for Artificial Intelligence.
The conference was quite a success, with almost 300 participants from an international audience and an impressive list of Keynote speakers. A lot of the contributions were about edge AI, which has a lot to do with NEUROTECH's remit, since low power consumption and event-based communication are two hallmarks of neuromorphic technology.
The NEUROTECH consortium committed to encourage a demo session at this conference by sponsoring a prize for the best demonstration of neuromorphic technology. Projects were evaluated by NEUROTECH members Yulia Sandamirskaya (INI Zürich) and Paolo Bortolotti (THALES), with support from Tobi Delbrück (INI Zürich).
Here are the winning projects:
"Flyintel – a Platform for Robot Navigation based on a Brain-Inspired Spiking Neural Network"
Huang-Yu Yao, Hsuan-Pei Huang, Yu-Chi Huang, and Chung-Chuan Lo, from National Tsing Hua University, Hsinchu, Taiwan
They showed network, inspired by circuits found in the fly brain for tracking the orientation of the insect, that was realised in a low-power spiking network simulation. The network was controlling and tracking the orientation of a robot, therefore demonstrating an “embodied” spiking network in a real-world environment. The winning contribution is that it demonstrated how biological inspiration can allow us to use very simple circuits — inspired by the smallest animals — to control robots efficiently.
“Artificial Intelligence of Things Wearable System for Cardiac Disease Detection”
Yu-Jin Lin, Chen-Wei Chuang, Chun-Yueh Yen, Sheng-Hsin Huang, Peng-Wei Huang, Ju-Yi Chen, and Shuenn-Yuh Lee, from National Cheng Kung University, Tainan, Taiwan
They have shown a working demonstration of a system for detecting cardiac diseases (one out of four) based on measurements from a single electrode. One could attach the electrode “live”, see recorded signals and the output of a CNN that was running on a PC/GPU (not on the device itself, but they have created the whole system, in which the electrode communicated the measurements wirelessly to a little box with the computer running CNN, with a nice GUI that showed the results). The demo was showing the prototype (which is close to clinical testing) in live operation. The neuromorphic aspects is brought in by the use of a state of the art neural network architecture to classify time series. Further, the use of neuromorphic hardware might allow to shed the big computing box that is still necessary to achieve real-time performance, therefore this application might be a good case promoting neuromorphic technology.
Congratulations and well done! The proud winning teams received well-earned 250 € each that hopefully inspire them to continue their great work and push forward the field of brain-inspired neuromorphic computing, and develop cutting-edge application based on this emerging technology.