Big data, machine learning pipelines for personalized display ads @Criteo

Gemini Solutions (www.geminisols.com) is glad to invite you to the technical presentation: Big data, machine learning pipelines for personalized display ads @Criteo.

Criteo is one of the largest players in the market of internet display ads. We have over 15000 partners - from verticals such as retail, travel or classified ads - who rely on us to deliver their ads to the right user, at the right moment and with the right content. In order to do this we rely on an ML system - the Criteo Engine - that contains two key components: the first is called Predictive Bidding and it refers to leveraging shopper data in order to bid for ad space on publisher sites at the right price; the second is called Product Recommendations and it refers to using shopper behavior to choose the items that are shown in the ads so that users are more likely to convert.

This presentation describes the pipelines we use to support these two components: offline to gather colossal amounts of data and feed it into our ML algorithms, online to bid for ads and compute product recommendations. We'll see what data we collect and how we centralize it in one of the largest Hadoop clusters in Europe. We'll then explore the life cycle of a display ad and see what Criteo components are used at various stages, with a particular focus on bidding and recommendation.

Free entrance! Food and drinks included!