Building Planetary-Scale Collaborative Intelligence

Pretrained models and cheap compute have made machine learning (ML) easier to deploy than ever, with the key bottleneck now being high-quality, relevant data. The data that is most valuable to decision-making is often distributed across networks of people and organizations, and is locked away in inaccessible silos due to unfavorable incentives and ethical-legal restrictions. This is starkly evident in healthcare, where such barriers have led to highly biased and underperforming tools.

In my talk, I will describe how collaborative systems (such as federated learning) provide a natural solution. They can bring together and utilize distributed data, while still respecting the agency, privacy, and interests of the data providers. Yet, for these systems to truly succeed, we must confront three fundamental challenges. These systems need to i) be efficient and scale to large networks, ii) provide reliable and trustworthy training and predictions, and iii) manage the divergent goals and interests of the participants. We discuss how tools from optimization, statistics, and economics can be leveraged to address these challenges.

Post Talk Link:  Click Here 

Passcode: H+WF!=2f