Research and Innovation at Featurespace

Our mission is to develop new algorithms and methodologies that make the world a safer place to transact. Our research continues the direction set by our academic founders from the University of Cambridge, Professor Bill Fitzgerald and his PhD student Dave Excell. We believe it is only through continued research and innovation that financial institutions can stay ahead of the criminals, and so we have set up our Innovation Lab in Cambridge (UK) to lead this effort on the behalf of the wider financial industry and its consumers.


Black Background

Deep Learning on Behavioral Sequences

Deep learning produces the most predictively powerful models, when its architecture is well tuned to the structure of the data. Our research focuses on applying sequential deep learning algorithms, such as recurrent neural networks and transformers, to behavioral event sequences. This research line underpinned both our launch of Automated Deep Behavioral Networks in 2021, and the launch of Tallier, the world's first Large Transaction Model, in Fall 2023.

Hero Image – TallierLTM – 2500×1700

TallierLTM™ Introducing the world’s first Large Transaction Model, created for data scientists, to protect consumers.

Learn more
Generative AI for financial transactions – photo-1608286022625-bc07f7a21154

Generative AI for financial transactions

Read article
Piotr Skalski 588_edited

Accelerating neural network training with large and precise gradient steps

Read More
Milly Farouki1269

Retraining Deep Neural Networks for Real-Time Transaction Processing

Learn more

Featurespace invents next generation of Machine Learning

Learn More
Automated Deep Behavioral Networks Featurespace

Automated Deep Behavioral Networks

Learn more

Privacy Enhancing Technologies

A lot of predictively useful information for fighting financial crime is siloed across organizations. This research line aims to create collaborative AI systems that share insights while preserving the privacy of sensitive data fields. This work is supported by a grant from Innovate UK, and received a prize in the PETs Prize Challenge Competition (2023).


Privacy-enhancing technologies prize challenges

Read More

How to combat financial crime through data sharing and collaboration

Read More

Trustworthy AI

Our systems make decisions that impact real lives, so it is imperative our systems can be trusted with those decisions. Our research into trustworthy AI aims to reinforce this trust for cutting-edge machine learning models by making their logic and impact transparent, interpretable, and fair.

Banner – How to make artificial intelligence articulate doubt -900pxX620px

How to make artificial intelligence articulate doubt

Learn more

Model Fairness in Fraud Prevention – Human Intervention and Practical Solutions

Read More
Iker Perez 345

Model Fairness in Fraud Prevention – Model Bias 101

Learn more
Web Header Cambridge Sessions

Cambridge Sessions

Watch Video
Black Background

The Future of AI Model Governance in Banking

Mission Critical Machine Learning

Latest News Visit the Newsroom