Split Learning: Distributed Machine Learning with Sensitive, Siloed Data

Institute of Microengineering - Distinguished Lecture (Bonus Lecture)

Abstract: Friction in data sharing is a large challenge for large scale machine learning.  Emerging technologies in domains such as biomedicine, health and finance benefit from distributed deep learning methods which can allow multiple entities to train a deep neural network without requiring data sharing or resource aggregation at one single place. The talk will explore the main challenges in data friction that make capture, analysis and deployment of ML. The challenges include siloed and unstructured data, privacy and regulation of data sharing and incentive models for data transparent ecosystems. The talk will compare distributed deep learning methods of federated learning and split learning. Our team at MIT has pioneered a range of approaches including automated machine learning (AutoML), privacy preserving machine learning (PrivateML) and intrinsic as well as extrinsic data valuation (Data Markets). One of the programs at MIT aims to create a standard for data transparent ecosystems that can simultaneously address the privacy and utility of data.

Bio: Ramesh Raskar is an Associate Professor at MIT Media Lab and directs the Camera Culture research group. His focus is on AI and Imaging for health and sustainability. They span research in physical (e.g., sensors, health-tech), digital (e.g., automated and privacy-aware machine learning) and global (e.g., geomaps, autonomous mobility) domains. He received the Lemelson Award (2016), ACM SIGGRAPH Achievement Award (2017), DARPA Young Faculty Award (2009), Alfred P. Sloan Research Fellowship (2009), TR100 Award from MIT Technology Review (2004) and Global Indus Technovator Award (2003). He has worked on special research projects at Google [X], Apple Privacy Team and Facebook and co-founded/advised several companies. Projet page https://splitlearning.github.io/

This lecture is part of the IMT Distinguished Lecture Series. The lecture is considered as a bonus lecture for the class MICRO-626 (usual attendance requirement does not apply, but participation is highly encouraged).