Our Technology

Our deep learning technology, Alchemite™, originated in work of Dr Gareth Conduit and his collaborators at the Cavendish Laboratory – the Physics Department of the University of Cambridge.

The Intellegens team further developed the technology and continues to push it forward, based on the requirements of our customers.

Alchemite™ trains deep neural networks to provide a unique combination of capabilities.

READ KEY PUBLICATIONS

A unique combination

Value from sparse, noisy data

A self-consistent, iterative algorithm imputes sparse data. This enables Alchemite™ to solve real-world problems that are not accessible to traditional machine learning.

Start fast

Alchemite™ generates a model from the available data that you can then refine and apply. There is minimal need for data cleaning and model setup, removing the high dependence on starting assumptions that is typical of many methods.

Unconstrained learning

Alchemite™ works across all available information to extract useful knowledge. This supports a holistic view (e.g., considering both ingredients and processing for a formulation), finding relationships that humans may miss.

Rational decisions

Alchemite™ offers advanced uncertainty quantification based on actual probability distributions for many properties simultaneously. This enables you to identify the experimental route most likely to succeed and to focus your time and effort based on a rational business case.

Optimise multiple targets

Alchemite™ can simultaneously predict multiple targets for high-dimensional data – an essential feature for formulation, discovery, and processing problems that is now available, even when the training data is sparse and noisy.

Scale up to large problems

Computations have a light memory and CPU footprint, so not only is Alchemite™ lightning fast, it can handle huge databases (where necessary).