Applied Brain Research Inc.

Applied Brain Research (ABR) produces AI tools, algorithms and applications for processing time-varying AI workloads for use in automotive, drone, phone, sensor, camera and autonomous-control applications. As an example, an ABR algorithm called the Legendre Memory Unit (LMU) makes temporal AI processing much more efficient.

Speech, video, health sensor, motor control, stock market and other AI signal processing workloads use far less power and memory while computing faster with the same or better accuracy than traditional AI methods. ABR has developed chip and software versions of the LMU. ABR's AI builder is Nengo, a visual, real-time, multi-platform AI compiler, used to build low-power, real-time AIs.

Nengo is a full-featured neural development environment that imports TensorFlow and Keras models. Nengo helps optimize models so they use 2x to more than 100x less power and can be deployed to CPUs, GPUs, FPGA, edge-chips and emerging neuromorphic chips. ABR is targeting device makers (car, phone, smart speaker, drone, camera, sensor, robot,) seeking to save power, extend battery life, compute more AI using smaller processors and less memory. We are also targeting cloud providers running automated speech recognition and/or video AI processing looking to cut 40% of their electricity bill on these workloads.

ABR is interested in meeting telecommunications customers and system integrators active in the telecommunications business.