Posted by Rob Knies

Eric HorvitzThis year’s Conference on Knowledge Discovery and Data Mining (KDD 2014) is themed “data science for social good.” That focus for the 20th KDD meeting moved Eric Horvitz (@erichorvitz) to accept an invitation to deliver a keynote address during the event, given his long interest in using machine intelligence to enhance the lives of people.

Horvitz, a Microsoft distinguished scientist and managing director of Microsoft Research Redmond, delivered an inspiring talk on Aug. 26 at the Sheraton New York Times Square Hotel—see his presentation here—before an audience of researchers and practitioners from the fields of data science, data mining, knowledge discovery, large-scale data analytics, and big data. His talk, on Data, Predictions, and Decisions in Support of People and Society resonated deeply with the core theme of the meeting.

It was an ideal setting for Horvitz, whose key interest lies in theoretical and practical challenges with developing systems that perceive, learn, and reason.

“When they invited me to do the talk,” he said, “I was really excited to find out that the meeting has a theme this year, Data Science for Social Good. This goal is near and dear to my heart, the idea of leveraging data as well as machine learning and decision-making—machine intelligence, more broadly—to contribute to humanity.

“As a young boy, I had hoped that I might one day be able to contribute to advancing science to help humanity. It’s been a lifelong passion, one that pervaded my grad-school work and on to a startup and then my work at Microsoft Research: How can we leverage these discoveries to help people as individuals, and society more broadly?”

Horvitz’s talk delved into advancements made in recent years in machine learning and decision support, and he sketched a broad, rich set of examples that ranged from deep learning and database analytics to traffic predictions, wind-monitoring systems, decision-making in health care, and community sensing.

“We published a set of principles and methods for the altruistic and privacy-sensitive use of data a few years ago called Toward Community Sensing,” he mentioned. “How could people ideally give such that everybody gets back? And how could systems be made keenly aware of preferences on giving?”

Horvitz, who holds both a Ph.D. in computing and an M.D. from Stanford University, has been passionate about predictive analytics and decision support in health care, and he shared details of several efforts in this realm in the KDD talk.

He discussed work in predicting hospital readmissions in the context of developing an informational pipeline that uses data to make a prediction that leads to a decision. He talked about the value of work to minimize errors in health care and thereby reduce the number of adverse outcomes and deaths. And he outlined efforts to use the World Wide Web as a massive set of sensors for planetary-scale analysis—and how such a system could help identify rare but serious adverse effects of medication.

On a broader scale, Horvitz also told his audience about research to apply machine learning and data-centric methods to enhance the quality of life worldwide, especially in developing countries. He cited his collaboration with colleagues on using existing infrastructure to understand, interpret, and respond to an earthquake—by measuring the rate of increase in mobile-phone usage after a quake strikes. That collaboration led to a nonprofit effort to pool data and stimulate research in machine intelligence for challenges in developing countries.

“There’s a great opportunity to help people out—and culture and society out—in a large variety of ways that might not be obvious to people who might be focusing more intensively on data science in the enterprise or for consumer services these days,” Horvitz said. “There’s so much focus on the latest smartphone application and services and recommender systems and better search results that we don’t think enough about the deep application of these methods to enhancing the quality of life in essential ways, including for people living in poverty and threatened by illness and hunger in lesser-developed countries.”