The Biggest Question in AI –

What Comes Next?

Business, success and entrepreneurship are about predicting the future and exploiting opportunities by driving towards them.

AI research page hero elements
element 1

Pioneering Advanced AI Research and Innovation

Our Research Themes Shaping Up Broad AI Models
Atomic-AI-icon.svg
Atomic AI
We used an ensemble of techniques to create AI models that are 50 times smaller. Additionally, we developed techniques to reduce training time and effort by over 50% and memory usage by over 90%.
Reliable-Vision-icon.svg
Reliable Vision
We have developed AI models for computer vision that perform effectively in the presence of translation, rotation, deformation, reflection, colour variations, scaling, and different orientations.
Neuro-Symbolic-AI-icon.svg
Neuro Symbolic AI
Combining symbolic logic with neural networks could be a promising path to Artificial General Intelligence (AGI). We have integrated logical reasoning into neural networks, using them to tackle problems in medical imaging, multi-modal diagnostics, and more.
Quantum-Inspired-AI-icon.svg
Quantum Inspired AI
By applying quantum computing principles to neural networks, we’ve developed a new model using tensor networks. This approach has allowed us to create AI models that are both highly accurate and compact, significantly enhancing their power and efficiency.
Neural-Turing-Machines-icon
Neural Turing Machines
We have implemented NTMs with von Neumann memory architectures to solve complex sequencing and planning problems.
Vector-Computing-icon.svg
Vector Computing 
This new AI approach uses large vector representations for cognitive tasks like reasoning, learning, and memory. It offers a more reliable, predictable, explainable, and robust method for AI development, which we apply to drug design, population health, and more.
Robust-AI-icon.svg
Robust AI
We have developed AI models that are robust and stable against adversarial attacks, making them more suitable for citizen-facing deployments.
Robust-AI-icon
Spiking Neural Networks & Neuromorphic Computing
We are at an advanced stage in developing efficient techniques for training spiking neural networks (SNNs) that are better and more practical than backpropagation through time.
artificial-general-intelligence
artificial general intelligence badge
The Quest For AGI
Have an AI project in mind?
Want us to work on your idea?

Hear From Our Happy Clients

5-stars
5 STAR REVIEWS ON
clutch-icon