What is recommendation? Personalized recommendation is the process of ranking a large collection of items based on users’ preferences. Accurate and efficient ranking of items enables providing personalized search results, marketing strategies, e-commerce production suggestions, and entertainment content.

Where is neural recommendation used? DNN-based personalized recommendation models are at the heart of many Internet services including search, social media, and online entertainment. These models also consume a significant fraction of computation cycles and infrastructure capacity devoted to AI in the datacenter and pose unique at-scale system design challenges for efficient training and inference.

Recommendation models consume the majority (over 80%) of AI inference cycles spent in Facebook’s datacenters.

Why do we need a tutorial on neural recommender systems? Despite the importance of DNN-based personalized recommendation models and their unique compute demands, relatively little research attention has been devoted to recommendation systems. To address this gap in the research community, this tutorial serves as a venue to discuss, share, and foster research for personalized recommendation systems. 


Tutorial Goals

  1. Providing historical context into the evolution of recommendation systems
  2. Discussing challenges of state-of-the-art recommendation systems
  3. Providing a hands-on tutorial on open-source benchmarks and datasets (training and inference)
  4. Brainstorming novel solutions for efficient personalized recommendation.


8:00am – 8:15amIntroduction to Neural RecommendationUdit Gupta and Carole-Jean Wu (Harvard and Facebook)
8:15am – 8:50amTraining Recommendation SystemsMaxim Naumov (Facebook)
8:50am – 9:25amDatasets for Recommendation SystemsFlavian Vasile (Criteo Labs)
9:25am – 10:00amAt-scale Inference for Recommendation SystemsUdit Gupta (Harvard University/Facebook)
10:00am – 10:30amCoffee Break
10:30am – 11:05amBuilding Production-Ready Recommendation Systems at Scale with MicrosoftAndreas Argyriou (Microsoft)
11:05am – 11:40amDesign Implications of Memory Systems and Near-Memory Processing for personalized recommendationXuan (Silvia) Zhang (Washington University/Facebook)

Tutorial Venue


Udit Gupta is a fourth year PhD student at Harvard University studying hardware specialization for AI and in particular, DNN-based personalized recommendation models. He is the leading authors of the “The Architectural Implications of Facebook’s DNN-based Personalized Recommendation” that presents the unique systems challenges of deploying deep learning based personalized recommendation at-scale.
David Brooks received his B.S. from University of Southern California in EE in 1997 and his M.A. and Ph.D. in EE from Princeton in 2001. He spent a year at IBM T.J. Watson Research Center in 2001 before joining Harvard in 2002. His research focuses on the interaction between the architecture and software of computer systems and underlying hardware implementation challenges, including power, reliability, and variability issues across embedded and high-performance computing systems. 
Gu-Yeon Wei received his B.S.E.E., M.S., and Ph.D. in Electrical Engineering from Stanford University in 1994, 1997, and 2001, respectively. In 2000, he joined Accelerant Networks (now a part of Synopsys) in Beaverton, Oregon as a Senior Design Engineer. In 2002, he joined Harvard University. His research interests span a variety of topics such as integrated voltage regulators, flexible voltage stacking, power electronics, low-power computing architectures and circuits, auto-parallelizing compilers, and more.
Brandon Reagen is an Assistant Professor in the Department of Electrical and Computer Engineering with affiliation appointments in the Computer Science. A computer architect by training, Brandon has a research focus on designing specialized hardware accelerators for applications including deep learning and privacy preserving computation. He has made several contributions to ease the use accelerators as general architectural constructs including benchmarking, simulation infrastructure, and System on a Chip (SoC) design. He earned a PhD in computer science from Harvard in 2018 and received his undergraduate degrees in computer systems engineering and applied mathematics from the University of Massachusetts, Amherst, in 2012.
Carole-Jean Wu is a Research Scientist at Facebook’s AI Infrastructure Research. She is also a tenured Associate Professor at Arizona State University. Carole-Jean’s research focuses in Computer and System Architectures. More recently, her research has pivoted into designing systems for machine learning. She is the leading author of “Machine Learning at Facebook: Understanding Inference at the Edge” that presents unique design challenges faced when deploying ML solutions at scale to the edge. Carole-Jean received her M.A. and Ph.D. from Princeton and B.Sc. from Cornell.   

Contact Us

Please contact us at: info.personal.tutorial@gmail.com