Vector-symbolic architectures as an algebraic programming language
for neuromorphic computers

Making the AI computations fast and more energy efficient is the current grand challenge of
Computer Science and Electrical Engineering. Solving it is essential for more rapid emergence of
intelligent technologies such as self-driving cars, autonomous robots, large scale information
retrieval systems. At the present time modern AI applications expose fundamental limitations of
the current computers.

To solve this challenge novel computing hardware such as neuromorphic
processors is emerging. The neuromorphic (NM) computers consume only a fraction of
the energy compared to the current technology and for certain tasks can be 1,000 times more
energy efficient. The purpose of this project is to contribute to the solution of the great AI
challenge by solving the challenges of the neuromorphic computing. With this project I aim at
bridging the NM computing usability gap by developing an algebraic programming methodology
for intuitive realization of AI models on the neuromorphic hardware; and showcase it by
developing novel AI algorithms tailored for efficient execution on NM hardware.

Funder: Swedish Research Council. Budget 4 600 000 SEK