Announcing the Criteo AI Lab

I am pleased to announce the launch of the Criteo AI Lab (CAIL), a new chapter in our journey on improving performance in computational advertising.

CAIL joins together Criteo Research with the Machine Learning Platform Engineering team, bringing together two main arms of R&D at Criteo. Typical configurations of research labs treat it as a stand-alone group whose sole mission is to publish papers, or embedded teams that help work on tactical problems. CAIL stands apart as a research organisation with the engineering power to develop and integrate new and state-of-art AI research into our production systems as well as to push forward our common understanding of AI techniques. With an investment of  e20M over three years, CAIL walks the talk of an AI lab that is central to the company’s mission.

The establishment of CAIL is timely as Criteo enters the paradigm of fundamentally understanding the relationship of a user with the products and brands they interact with. At 1.4B monthly shoppers interacting with a few billion products across 18,000 partners across multiple countries, personalization has always been a challenging problem. However, with recent advancement in machine learning such as deep learning architectures from other domains, the timing is right to investigate taking our predictive performance to the next level. We expect our research to improve our existing products as well as create capabilities than could benefit products we don’t imagine yet.

Much ink (and online pixels) have been spilled by the yay and naysayers on the state of AI and in particular the potential of deep models. CAIL will take a more nuanced stance, which we believe should be the hallmark of any research team. We want to fundamentally understand why models work the way they do, understand their ability to generalize, test their fragility to push forward our understanding the technology. We remain true to our research roots and also report on baselines that don’t oversell the dream. To this end, making AI interpretable will be a core mission of the group.

CAIL will continue the Criteo tradition of fostering our connections with the academic community via our Faculty Research Program, internships, sabbaticals and support of PhD students. We will continue to fully participate in the vibrant AI community by publishing our research in conferences, open-sourcing software, contributing to online ML repositories and by continuing to release datasets that can serve as benchmarks. Needless to say, we remain committed to making sure our research confirms to the highest ethical standards.

We are excited to start on our new journey and look forward to continuing to transform every ad into a relevant shopping experience for our users.

Suju Rajan
VP, Head of Research
CAIL