Date

2020.7

Federated Recommender System

Overview

The recommender system (RecSys) plays an important role in the real-world applications, from product recommendations to news recommendations. It has become an indispensable tool for coping with information overload. In general, RecSys is heavily data driven, the more data is used in the Recsys, the better recommendation performance can be obtained. They collect the private user data, such as the behavioral information, contextual information, the item metadata, the purchase history, the recommendation feedback, the social information and so on from different data sources. All this informative user data is centrally stored at the database of each organization to support different kinds of recommendation services.

However, collecting data from multiple parties could lead to serious privacy and security risks and violate laws such as the General Data Protection Regulation (GDPR). In this tutorial, we introduce a new notion of Federated Recommender System (FedRec). Compared to the conventional RecSys, FedRec protects the user privacy and data security through decentralizing private user data locally on each party. In this tutorial, first of all, according to the data structure of recommendation tasks, we categorize the FedRec into horizontal FedRec Systems, vertical FedRec systems and transfer FedRec systems. We explain the problems and typical scenarios in each category. Secondly, we implement an open-source tool that contains typical algorithms in each category. The tool is based on FATE, which is an industrial level framework designed to support federated learning architectures and secure computation. We demonstrate the idea and implantation details of each FedRec algorithm. Thirdly, we show two applications in news recommendation and online advertising. Finally, as a promising direction with huge potential opportunities, we discuss challenges and open questions in FedRec systems.

Outline

Introduction

Part I:  Theory and Algorithms of FedRec Systems (~1:45hour)

1) Preliminary of Recommendation (~20min)

2) Preliminary of Federated Learning (~30mins)

3) Categorization and typical algorithms (~50mins)

Part II:  Implementation and Applications of FedRec Systems (~1:45hour)

4) FATE: open-source project to support the federated AI ecosystem (~20mins)

5) FedRec algorithms and their implementations (~50 mins)

6) Applications and future works (~35 mins)

Materials

Slides in PDF（Upcoming stay tuned）

Open-sourced Project

Demo

Presenters

·Qiang Yang: Chair Professor at Department of Computer Science and Engineering, Hong Kong University of Science and Technology; Chief AI Officer, WeBank, China

Prof. Yang is currently a Chair Professor in the Department of Computer Science and Engineering of the Hong Kong University of Science and Technology, and Chief AI Officer at WeBank, China. He received his MSc degree in Astrophysics, MSc in Computer Science and PhD in Computer Science at University of Maryland, USA.

His research interests include federated learning, transfer learning, machine learning, planning and data mining in artificial intelligence. He is also the President of IJCAI, Executive Council Member of AAAI, and Editor in Chief of IEEE Transactions on Big Data. He was the founding Editor in Chief of ACM Transactions on Intelligent Systems and Technology. He was the founding director of the HKUST’s Big Data Institute, the founding director of the Huawei Noah Ark Research Lab, and the founding director of Wechat-HKUST Joint Lab on AI. He is a fellow of ACM, IEEE, AAAI, AAAS, IAPR and CAAI.

·Ben Tan: Senior researcher in the AI department of WeBank, China

Dr. Tan is currently a senior researcher in the AI department of WeBank, China. Previously, he was a senior researcher at Recommendation Center, Tencent, China. He received his PhD in Computer Science at the Hong Kong University of Science and Technology.

His research interests include Recommender systems, online advertising, federated learning, transfer learning. He has over five years experiences on recommender systems and online advertising. He has published over ten papers on several top-tier conferences and journals such as KDD, AAAI, SDM, Pattern Recognition, IEEE Transactions on ITS, ACM Transactions on IST, and presented his work in multiple external venues. He serves on the program committee member of WWW, IJCAI, CIKM, SDM and related conferences.

 

Contact Us:

Email: ai.ads@webank.com

Official Stie: https://ad.webank.com/