Neuro-symbolic AI on Cirrus High-Performance Compute

UMNAI collaborates with EPCC to advance our novel Neuro-symbolic AI and optimise it for high-performance compute (HPC) environments like EPCC’s Cirrus.

In an industry-academia collaboration with the University of Edinburgh’s EPCC, UMNAI is deploying our novel Neuro-Symbolic AI algorithms and our model induction (generation) technology on EPCC’s Cirrus High-Performance Compute (HPC) system.

This collaboration will prove the efficacy of UMNAI’s algorithms and induction process in distributed training environments and with very large datasets. It will also serve to optimise and test overall data efficiency of UMNAI’s technology.

UMNAI’s Neuro-symbolic AI forms the foundation of our Hybrid Intelligence Framework, a system that enables the integration of human knowledge with explainable AI in a collaborative manner. The framework empowers decision workflows that apply intentional reasoning and oversight to ensure fit-for-purpose decisions. In Hybrid Intelligence powered decision workflows, human crafted tests, measurements and assessments intelligently interrogate each prediction to fully understand it. With awareness and understanding, decisions are programmatically aligned to meet the intentions and obligations of the operator.

UMNAI’s Neuro-symbolic AI and Hybrid Intelligence Framework are particularly suited to the application of AI in regulated industries such as financial services, insurance, healthcare, social services and other highly regulated industries.

Contact us if you have an application that could benefit from UMNAI’s technology or if you’d like to learn more about the benefits of our Neuro-Symbolic AI.


Cirrus is one of the EPSRC Tier-2 National HPC Facilities and is available for users in both academia and industry.

Housed at EPCC’s Advanced Computing Facility at the University of Edinburgh, Cirrus is a flexible, state-of-the-art High Performance Computing system that provides an ideal platform for users to solve their computational, simulation, modelling, and data science challenges.

The facility is based around a 280 node HPE/SGI 8600 HPC system with a DDN Lustre file system. The HPC system runs the CentOS Linux distribution and has a range of software, libraries and tools available.

Find out more about Cirrus