Author | AIConIn the real problems of "data islands" and "privacy protection", using data to continuously optimize the effects and provide high-quality services under the premise of reasonableness and lawfulness is the huge challenge and primary task faced by the current recommendation system.
The recommendation system is widely used and has been used in all aspects of people's lives, such as news recommendation, video recommendation, product recommendation, etc. In order to achieve accurate recommendations, the recommendation system collects data on a large number of users and recommended content. Generally speaking, the more data collected, the more comprehensive and in-depth understanding of users and recommended content, and the more accurate the recommendation effect. In reality, as user data security and privacy protection related policies are successively promulgated and improved day by day, these data are usually scattered in different organizations in the form of "data islands" in order to protect user data privacy. Therefore, in the real problems of "data islands" and "privacy protection", using data under the premise of reasonable and legal conditions to continuously optimize the effect to provide high-quality services is a huge challenge and primary task faced by the current recommendation system.
At the AICon Shanghai 2020 conference, Tan Ben, a senior researcher at WeBank, will introduce a new federal recommendation system, which uses data legally and reasonably while protecting the privacy and data security of partners, significantly improving the recommendation effect.
Content outline
1. Introduction to the background of the recommendation system
Recommend system introduction
Privacy issues in recommender systems
2. Federal Recommendation System
Federated learning
Federal recommendation system and its classification
Federated recommendation algorithm
3. Architecture of the WeChat Federated Recommendation System
FATE Federal Learning Framework
Federated recommendation system architecture
4. Application of Federal Recommendation System in Financial Marketing
Federal content recommendation
Federal Online Advertising
Audience benefits
- Understand the latest progress of federated learning in the recommendation system;
- Understand the core concepts and technical principles of the federal recommended technology;
- Understand the implementation of federal recommendation technology in financial marketing.
Highlights of the speech
Paying attention to data privacy and security has become a worldwide trend, and "federated learning" is a key technology to solve this industry-specific problem. For the first time, WeBank has applied federated learning technology to the practice of recommendation in the field.
Suitable for the crowd
Participants have a good understanding of one or two areas of recommendation systems, federated learning, machine learning, and data security.
Lecturer profile
Tan Ben is a senior researcher at WeBank. He graduated with a Ph.D. from the Hong Kong University of Science and Technology. His main research directions are migration learning, recommendation systems, machine learning, etc. He is currently responsible for the research and development of recommendation algorithms in the artificial intelligence department of WeBank. He once worked at Tencent and was responsible Advertising conversion rate estimation, product recommendation, etc. He has published more than 10 papers in international academic journal conferences such as KDD, AAAI, SDM, TIST, and has served as a member of the program committees of international conferences such as WWW, IJCAI, CIKM, SDM, etc., and has repeatedly ranked among the best in international data mining competitions.