NextSmarty’s technology is in Criteo’s top 10

Industry leader features NextSmarty

Criteo is a world leader in digital advertisement and retargeting marketing, and we are proud to have our research publication in the top 10 list of papers selected by Criteo. 

They selected our work as we took part at the ACM RecSys 2017 conference in Como (Italy). The RecSys is one of the most important events in our field during the year, featuring over 600 participants from academia and industry from all over the world, and that gives an idea of how flattered we are to be spotted by one of the main sponsors of the event.

“This work opens up the field of sequence modeling for recommendation to convolution models that have the potential of being faster to train and easier to understand.”

They said that our engine has a high potential: we couldn’t agree more. Innovation is our keyword and we can’t wait to show how further we’ve brought our studies and how good the results that we achieved with our customers are. 

Criteo’s review of NextSmarty’s research publication

Other research publications featured in Criteo’s top 10

Here’s a quick overview of some of the other studies reviews:

  • Yahoo Labs

“Expediting Exploration by Attribute-to-Feature Mapping for Cold-Start Recommendations”  by Yahoo Labs, a  very interesting study concerning the cold start problems, as for contents with no interaction history. The authors propose a way to map an item’s latent feature adopting the attribute-to-feature approach, and numbers show how this “learning technique achieves more accurate initial estimates than logistic regression methods.

  • Research at Google

Folding: Why Good Models Sometimes Make Spurious Recommendations by Research at Google, they face the problem of spurious recommendation by investigating folding: “the unintentional overlap of disparate groups of users and items in the low-rank embedding vector space, induced by improper handling of missing data.”

  • Telefonica

Personalizing Session-based Recommendations with Hierarchical Recurrent Neural Networks by Telefonica, the authors propose a “way to personalize RNN models with cross-session information transfer and devise a Hierarchical RNN model that relays end evolves latent hidden states of the RNNs across user sessions.”

Among the sponsors of those finest studies, there are universities and private research institutes such as Politecnico di Milano, UC San Diego, Google Research, Yahoo Labs, and Telefonica.

We’re glad to be in such a good company, and we would like to thank Criteo Labs for reviewing our research publication and for choosing it for their top 10.

Special thanks go also to the ACM organization that let us join this year’s conference, it was a wonderful experience filled with inspiration and very good talks about AI.

breitling navitimer
rolex oyster
cartier santos