WiMLDS x Criteo: ML in Computational Advertising


Join us in Palo Alto for our first night of tech talks in 2019! Listen as our speakers from Criteo share the unique challenges of machine learning in computational advertising. First, you'll learn from Suju about delivering real-time prediction at scale -- of hundreds of billions of requests per day! Next, Fengjiao will show you some techniques for consolidating messy, real-world product catalogs with machine learning. Lastly, Diane will showcase how Criteo's recommender system delivers recommendations on an extensive, ten billion item, product catalog in about 10ms.


6:30 Doors open
7:00 Welcome from WiMLDS, Criteo
7:05 Computational Advertising at Scale, Suju Rajan
7:35 Building a Universal Catalog, Fengjiao Wang
8:05 Recommendations at Scale at Criteo, Diane Gasselin
8:30 Networking - Food, drinks are served
9:30 Doors close


Computational Advertising at Scale, Suju Rajan

Abstract: Machine learning literature on Computational Advertising typically tends to focus on the simplistic CTR prediction problem which while being relevant is the tip of the iceberg in terms of the challenges in the field. There is also very little appreciation for the scale at which the real-time-bidding systems operate (200B bid requests/day) or the increasingly adversarial ecosystem all of which add a ton of constraints in terms of feasible solutions. In this talk, I’ll highlight some recent efforts in developing models that try to better encapsulate the journey of an ad from the first display to a user to the effect on an actual purchase.

Speaker Bio: Suju Rajan is the SVP, Head of the Criteo AI Lab. At Criteo, her team works on all aspects of performance driven computational advertising, including, real-time bidding, large-scale recommendation systems, auction theory, reinforcement learning, online experimentation, metrics and scalable optimization methods. Prior to Criteo, she was the Director of the Personalization Sciences at Yahoo Research where her team worked on personalized recommendations for several Yahoo products. She received her PhD from the University of Texas at Austin.

Building a Universal Catalog, Fengjiao Wang

Abstract: Real-world product catalogs are more complicated than one imagines them to be. Building machine learning solutions to bring together and normalize these different catalogs is a highly challenging machine learning problem. In this talk, I’ll introduce some of the challenges, the machine learning solutions we devised to tackle the same and the lessons we learned along the way.

Speaker Bio: Fengjiao Wang is a research scientist at the Criteo AI Lab. In her current role, she helps build machine learning solutions for various Criteo initiatives – spanning ML topics such as categorization, image similarity, product & user embedding & product disambiguation. Fengjiao holds a Ph.D. in Computer Science from UIC. In her free time, she likes to take her newborn baby to storytelling sessions in nearby libraries and reads picture books to her.

Recommendations at Scale at Criteo, Diane Gasselin

Abstract: Criteo displays online advertisements that are personalized according to each user's browsing history and optimized for post-click sales. The challenge of our recommender system is to choose a handful of relevant products among 10B of candidates in about 10ms. We face a trade-off between the amount of data we can ingest and the latency requirements of ad serving. Also, as we work with 20K+ clients that each have their catalog of products, a side challenge is to unify these products across the partners.

Speaker Bio: Diane Gasselin has been a Software Developer at Criteo for 5 years, working on various topics such as prediction, recommendation, and product classification. She worked both on data pipeline jobs on Hadoop and on web services to support our online display requests.