Abstract: Data is at the core of Machine Learning/Artificial Intelligence. Usually model performance improves with more data available. Often, data an organization has is not sufficient and needs to be augmented with additional information. Nevertheless, there are data privacy concerns when it comes to cross-organisation personal data sharing. There is an increasing awareness of staying compliant when sharing under regulations such as GDPR and PDPA. Even if it can be ensured that the personal data sharing stays compliant, there are plenty of business/commercial considerations that do not give enough justification/incentive for organisations to share data. Federated learning is a privacy-preserving machine learning technique that trains a model across multiple decentralised parties holding local data, without exchanging them.
AI Singapore has been working on building a system, named Synergos, to support Federated Learning. In this talk, we will present an overview of the key component of Synergos, and zoom into the core component which coordinates multiple parties to train a federated model.
In the talk, a couple of thing I'm going to cover:
1. FL 101, i.e. key concepts in FL.
2. the design of Synergos. Synergos is built on top of Pysyft. We will talk about how it is designed to support collaborative training without exposing individual data, which is key of FL. Synergos hides many low-level details of FL from end-users, so that FL can be made more user friendly.
3. we will also share a use case how synergos can be used, so that the attendees would see it in action in real setting. A demo will be given (if time permits)
Bio: Jianshu is Head of Federated Learning at AI Singapore where he leads team to develop a platform to support Federated Learning, a new paradigm of privacy-preserving Machine Learning. As a national initiative, AISG brings together the strength of AI research bodies in Singapore’s Autonomous Universities and research institutes, together with the vibrant ecosystem of AI start-ups and companies developing AI products, to perform use- inspired research, create innovative AI solutions, and develop the talent to power Singapore’s AI efforts.
Jianshu has many years of AI/Data Science research and consulting experience. In recent years, he has spent most of his time in putting AI/ML into real-world usage and promoting ethical aspects of AI/ML, e.g. explainability, fairness, robustness, and privacy- preserving AI/ML models. Before joining AISG, he was the Head of Insights and Modelling of a leading global reinsurer where he led his team to deliver a number of significant data science projects for their key clients in the Asia Pacific region..
Jianshu obtained his Ph.D in Computer Science from Nanyang Technological University (NTU), Singapore in 2008.