Sham M. Kakade

Harvard University

University of Washington

Senior Data Science Fellow, eScience Institute

Adjunct Professor in Department of Electrical Engineering

Sr. Principal Researcher at MSR-NYC

Research

I am interested in the mathematical foundations of machine learning and AI. I focus on the design of provably efficient and practical algorithms that are relevant for a broad range of paradigms. My current interests include : i) reinforcement learning and controls ii) representation (and deep) learning and iii) natural language processing and memory. I enjoy collaborating with a diverse set of researchers to tackle these issues!

Book

Reinforcement Learning: Theory and Algorithms

Alekh Agarwal, Nan Jiang, Wen Sun, and I are writing a monograph on Reinforcement Learning. We will be periodically making updates to the draft. Also, see current course CS 6789: Foundations of Reinforcement Learning.

WA Exposure Notification App

WA Governor’s Office

WA Governor’s Office

November 30, 2020

Inslee announces statewide COVID-19 exposure notification tool

WA Notify uses privacy-preserving technology and works without collecting or revealing any location or personal data

Washington Exposure Notifications - WA Notify

Washington Exposure Notifications - WA Notify

May 6, 2020
WA Notify (also known as Washington Exposure Notifications) is a free tool that works on smartphones to alert users if they may have been exposed to COVID-19 without sharing any personal information.  It is completely private and doesn’t know who you are or track where you go.

COVID19 Reports and CommonCircle

D. Siddarth, et al., Mitigate/Suppress/Maintain: Local Targets for Victory Over COVID. Edmond J. Safra Center for Ethics: Rapid Response Initiative, 2020. Publisher's VersionAbstract
There is growing consensus around a strategy centered on testing, tracing, and supported isolation (TTSI) to suppress COVID, save lives, and safely reopen the economy. Given the high prevalence the disease has reached in many OECD countries, this requires the expansion of TTSI inputs to scales and with a speed that have little precedent (Siddarth and Weyl, 2020). As scarcity of testing supplies is expected to remain during the build-up to a surge, authorities will need to allocate these tests within and across localities to minimize loss of life. This paper documents a preliminary framework that aims to provide such guidance to multiscale geographies, in conjunction with our previous recommendations. Given unfortunate limits in current testing capacity, we describe a testing and tracing regime in which orders of magnitude fewer resources can be used to suppress the disease. Such suppression should be rapidly scaled in low-prevalence areas (within and across regions) to the extent that those areas can be insulated from other areas. In particular, if travel restrictions are used to allow asynchronous sup-pression, and if logistics permit the use of mobile resources, a smaller number of tests per day may be required. At the same time, vulnerable communities and essential workforces must be protected across regions, as prescribed in Phase I of the Roadmap to Pandemic Resilience (Allen et al., 2020).
D. S. Allen, A. Bassuk, S. Block, G. J. Busenberg, and M. Charpignon, Pandemic Resilience: Getting it Done. Edmond J. Safra Center for Ethics: Rapid Response Initiative, 2020. Publisher's VersionAbstract
On April 27, the CDC [Centers for Disease Control and Prevention] changed its guidance to support broader use of testing not only for therapeutic purposes, but also for disease control. In the most recent guidance, released May 3, first priority goes to hospitalized patients, first responders with symptoms, and residents in congregate living contexts with symptoms. But there is now also a second priority category that includes asymptomatic individuals from groups experiencing disparate impacts of the disease and 'persons without symptoms who are prioritized by health departments or clinicians, for any reason, including but not limited to: public health monitoring, sentinel surveillance, or 'screening of other asymptomatic individuals according to state and local plans' (bold in original, italics added). The last phrase supports broad testing of contacts of COVID [coronavirus disease]-positive individuals and of essential workers, even when they have mild symptoms or none at all. This Supplement to our Roadmap to Pandemic Resilience offers guidance to help state and local governments develop TTSI (testing, tracing, and supported isolation) programs in support of such testing for purposes of disease control and suppression.
V. Hart, et al., Outpacing the Virus: Digital Response to Containing the Spread of COVID-19 while Mitigating Privacy Risks. Edmond J. Safra Center for Ethics: Rapid Response Initiative, 2020. Publisher's VersionAbstract
There is a growing consensus that we must use a combined strategy of medical and technological tools to provide us with response at a scale that can outpace the speed and proliferation of the SARS-CoV-2 virus. A process of identifying exposed individuals who have come into contact with diagnosed individuals, called “contact tracing,” has been shown to effectively enable suppression of new cases of SARS-CoV-2 (COVID-19). Important concerns around protecting patient’s confidentiality and civil liberties, and lack of familiarity with available privacy-protecting technologies, have both led to suboptimal privacy implementations and hindered adoption. This paper reviews the trade-offs of these methods, their techniques, the necessary rate of adoption, and critical security and privacy controls and concerns for an information system that can accelerate medical response. Proactive use of intentionally designed technology can support voluntary participation from the public toward the goals of smart testing, effective resource allocation, and relaxing some of physical distancing measures, but only when it guarantees and assures an individual’s complete control over disclosure, and use of data in the way that protects individual rights.
J. Chan, et al., “PACT: Privacy Sensitive Protocols and Mechanisms for Mobile Contact Tracing,” IEEE Bulletin on Data Engineering, pp. 15-35, 2020. Publisher's VersionAbstract
The global health threat from COVID-19 has been controlled in a number of instances by large-scale testing and contact tracing efforts. We created this document to suggest three functionalities on how we might best harness computing technologies to supporting the goals of public health organizations in minimizing morbidity and mortality associated with the spread of COVID-19, while protecting the civil liberties of individuals. In particular, this work advocates for a third-party free approach to assisted mobile contact tracing, because such an approach mitigates the security and privacy risks of requiring a trusted third party. We also explicitly consider the inferential risks involved in any contract tracing system, where any alert to a user could itself give rise to de-anonymizing information. More generally, we hope to participate in bringing together colleagues in industry, academia, and civil society to discuss and converge on ideas around a critical issue rising with attempts to mitigate the COVID-19 pandemic.

News

Bernardo Sabatini and Sham Kakade.

New University-wide institute to integrate natural, artificial intelligence

December 7, 2021
Harvard University on Tuesday launched the Kempner Institute for the Study of Natural and Artificial Intelligence, a new University-wide initiative standing at the intersection of neuroscience and artificial intelligence, seeking fundamental principles that underlie both human and machine intelligence. 

Activities and Services