Industry & Government Track Invited Talks
Please refer to the program page for the times and locations of these talks.

It is extremely important in many application domains to have transparency in predictive modeling. Domain experts do not tend to prefer "black box" predictive models. They would like to understand how predictions are made, and possibly, prefer models that emulate the way a human expert might make a decision, with a few important variables, and a clear convincing reason to make a particular prediction.

I will discuss recent work on interpretable predictive modeling with decision lists and sparse integer linear models. I will describe several approaches, including an algorithm based on discrete optimization, and an algorithm based on Bayesian analysis. I will show examples of interpretable models for stroke prediction in medical patients and prediction of violent crime in young people raised in out-of-home care.

Cynthia Rudin is an associate professor of statistics at the Massachusetts Institute of Technology associated with CSAIL and the Sloan School of Management, and directs the Prediction Analysis Lab. Previously, Prof. Rudin was an associate research scientist at the Center for Computational Learning Systems at Columbia University, and prior to that, an NSF postdoctoral research fellow at NYU. She holds an undergraduate degree from the University at Buffalo where she received the College of Arts and Sciences Outstanding Senior Award in Sciences and Mathematics, and she received a PhD in applied and computational mathematics from Princeton University in 2004. She is the recipient of the 2013 INFORMS Innovative Applications in Analytics Award. She was given an NSF CAREER award in 2011. Her work has been featured in IEEE Computer, Businessweek, The Wall Street Journal, the Boston Globe, the Times of London, Fox News (Fox & Friends), the Toronto Star, WIRED Science, Yahoo! Shine, U.S. News and World Report, Slashdot, CIO magazine, and on Boston Public Radio.

Petabytes of data about human movements, transactions and communication patterns are being generated by everyday technologies such as mobile phones and credit cards. This unprecedented volume of information facilitates a novel set of research questions applicable to a wide range of development issues. In collaboration involving 237 mobile phone operators across 102 countries, Jana’s mobile technology platform can instantly poll and compensate 3.48 billion active mobile subscriptions. This talk will discuss how insights gained from living in Kenya became the genesis of a technology company currently working with global clients in over 50 countries, including P&G, Google, Unilever, Danone, General Mills, Nestle, Johnson & Johnson, Microsoft, the World Bank, and the United Nations. After providing an overview of the mobile and social media landscapes in emerging markets, we discuss a system that implements polls and mobile subscription compensation. The presentation will conclude by emphasizing the value of consumer data in underserved and understudied regions of the world.

Nathan Eagle is the co-founder and CEO of Jana, a mobile technology platform that connects emerging market consumers with global brands using mobile airtime. Nathan is also an Adjunct Assistant Professor at Harvard University, where he researches how data generated by human movements, financial transactions, and communication patterns can be used for social good. He has been named to Advertising Age’s 40 under 40 in 2014 and MIT’s TR35, a group of the world’s top innovators under 35 in 2009. Nathan current serves on the World Economic Forum as Technology Pioneer. He holds a BS and two MS degrees from Stanford’s School of Engineering as well as a PhD from MIT.

E-commerce has largely been a “Pull” model to date. Offline retailers have nailed discovery, delight, serendipity, and impulse purchases in person with greater success than online commerce sites. However, in an always-on, mobile-first world, companies like Groupon have the opportunity to push the frontier even further than offline retailers or comprehensive sites due to the fact that our smartphones are always with us. The challenge is to provide the right deals to the right user at the right time. That involves learning about the users and their locations, their personal preferences, and predicting which deals are likely to delight them, presenting diversity, discovery and engaging UX to gather user preferences and semantic graph approaches for user-deal matching. This presentation will give insight into how Groupon manages to grapple with these challenges via a data-driven system in order to delight and surprise customers.

Sri Subramaniam is VP of Engineering for Relevance and Search (Smart Deals) at Groupon. His responsibility is to put the right deals in front of the right consumers at the right time, for mobile, web and e-mail. Subramaniam comes to Groupon from WalmartLabs, where he was VP of Engineering, responsible for technologies such as Search, Catalog, Marketplace and Item infrastructure. WalmartLabs was created through the acquisition of Kosmix, a Semantic Tech startup, where Sri was VP, Engineering. Prior to Kosmix, Sri worked at eBay, responsible for Search and other e-commerce systems. Sri is passionate about E-commerce, Search & Personalization, and Consumer Apps. He holds a B.Tech in Electronics from the Indian Institute of Technology, Madras, and an M.S. in Computer Engineering from Rensselaer Polytechnic Institute.

It’s not often we get to peer into the details of how big corporates use predictive modeling in practice. In this talk, Sprint’s Head of Predictive Modeling, Tracey De Poalo, will talk about the process she developed using SAS and logistic regression to build a wide range of models. Jeremy Howard will discuss his experience as a consultant at Sprint comparing R and random forests to the existing process, and will show the pros and cons of each approach.

The talk will also cover the “real world” issues that Tracey has dealt with in creating a reusable process, including data mart development, sampling, testing, reporting, and implementation with internal customers.

Compared to Jeremy’s 2013 talk, which was focused on algorithms, this is focused on process. Sprint’s existing process is the best that Jeremy has seen in industry, and shows a lot of best practices in data structure, documentation, testing, automation, and customer interaction, and modeling.

Tracey has been managing the Predictive Analytics team at Sprint for six years. Her educational background includes a Masters in Statistics from Florida State University and a Masters in Industrial Design from Georgia Institute of Technology. A deep understanding of design methodology allows Tracey to bring creative solutions to complex analytical problems. She has introduced multiple, robust procedures that have made predictive analytics at Sprint more efficient and productive. Tracey also has extensive analytical experience in the credit industry working for Equifax and Merkle.

Jeremy Howard is an entrepreneur, business strategist, developer, and educator. He is the CEO of Enlitic, which is bringing modern machine learning to medical diagnostics (with a particular focus on deep learning for medical imaging). He is also a faculty member at Singularity University, and a Young Global Leader with the World Economic Forum. He was a Distinguished Research Scientist at the University of San Francisco, and was the Data Strategist for Khosla Ventures. Previously he was the President and Chief Scientist of the data science platform Kaggle. He was the founding CEO of two successful Australian startups (FastMail, and Optimal Decisions Group). Before that, he spent 8 years in management consulting, at McKinsey & Co, and AT Kearney. Jeremy has invested in, mentored, and advised many startups, and contributed to many open source projects.

In the era of EHRs, it is possible to examine the outcomes of decisions made by doctors during clinical practice to identify patterns of care—generating evidence based on the collective practice of experts. We will discuss methods that use unstructured patient data to monitor for adverse drug events, profile specific drugs, identify off-label drug usage, uncover ‘natural experiments’ and generate practice-based evidence for difficult-to-test clinical hypotheses. We will describe how to detect associations among drugs and their adverse events several years before an alert is issued as well as compute the true rate of drug-drug interactions. We will present approaches to identify novel off-label uses of drugs using the patient feature matrix along with prior knowledge about drugs, diseases, and known usage. We will review a natural experiment—where a subset of CHF patients who were prescribed Cilostazol despite its black box warning—and profile its safety. We will discuss the testing of a clinical hypothesis about an association between allergic conditions and chronic uveitis in patients with juvenile idiopathic arthritis.

Dr. Nigam H. Shah is an Assistant Professor of Medicine (Biomedical Informatics) at the Stanford School of Medicine. Dr. Shah researches methods of using machine learning, text-mining, and medical ontologies to learn medical evidence from unstructured clinical data. He teaches a graduate class on data-driven medicine. Dr. Shah holds an MBBS from Baroda Medical College, India, a PhD from Penn State University, USA and completed post-doctoral training at the Stanford Medical School in 2007.

In this talk, Drew will examine data science through the lens of the social scientist. He will discuss how the various skills and disciplines combine into data science. Drew will also present a motivating example directly from his work as a senior advisor to NYC's Mayor's Office of Analytics.

Drew Conway is a leading expert in the application of computational methods to social and behavioral problems at large-scale. Drew is the is the Head of Data at Project Florida, and has been writing and speaking about the role of data — and the discipline of data science — in industry, government, and academia for several years.

Drew has advised and consulted companies across many industries; ranging from fledgling start-ups to Fortune 100 companies, as well as academic institutions and government agencies at all levels. Drew started his career in counter-terrorism as a computational social scientist in the U.S. intelligence community.

This presentation will cover how natural language processing and distributed human computing help improve the lives of speakers of all the world's languages, in applications including education, disaster-response, health and access to employment. When applying natural language processing to the full diversity of the world's communications, we need to go beyond simple keyword analysis and implement complex technologies that require human-in-the-loop processing to ensure usable accuracy. I will give examples from recent work where more than a million human judgments were collected on unstructured text and imagery data around natural disasters - the largest application of its kind by an order of magnitude. The observations debunk some of the over-optimistic claims about the utility of social media following disasters. On the positive side, I will share results that show how for-profit technologies are improving people's lives by providing sustainable economic growth opportunities when they support more languages, aligning business objectives with global diversity.

Rob is the CEO of Idibon, a company that provides languages technologies for all the world's languages. He has worked for the UN Refugee Commission in Liberia and Sierra Leone; for disaster response organizations in Haiti; and in tracking disease outbreaks globally. Outside of work, he has learned about the world's diversity by cycling more than 20,000 kilometers across 20 countries. Rob has a PhD from Stanford University.

The US Department of Defense Dictionary of Military terms defines the Information Environment as "the aggregate of individuals, organizations, and systems that collect, process, disseminate, or act on information." The decisions and actions we take both as individuals and collectively simultaneously shape and are shaped by the information environment in which we live. The nature of our interaction with the information environment is rapidly evolving and old models are becoming irrelevant faster than we can develop new ones. This results in uncertainties that leave us exposed to dangerous influences without proper defenses. The purpose of this talk is to help frame a new science of Information Environment Security (IES) whose goal is to create and apply the tools needed to discover and maintain fundamental models of our ever-changing information environment and to defend us in that environment, both as individuals and collectively, against intentional as well as unintentional attempts to deceive, misinform and otherwise manipulate us. IES is an interdisciplinary science that will require bringing together experts working in areas such as cognitive science, computer science, social science, security, marketing, political campaigning, public policy, and psychology to develop a theoretical as well as an applied engineering methodology for managing the full spectrum of information environment security issues.

Dr. Rand Waltzman joined DARPA as a Program Manager in May 2010. His focus is on information systems security.

Dr. Waltzman joined DARPA from Lockheed Martin Advanced Technology Laboratories (LM-ATL), where he served as Chief Scientist for the Applied Sciences Laboratory that specializes in advanced software techniques and the computational physics of materials. Prior to LM-ATL he was an Associate Professor in the Department of Computer Science at the Royal Institute of Technology in Stockholm, Sweden, where he taught and performed research in applications of machine intelligence technology to a variety of problem areas including digital entertainment, automated reasoning and decision support and cyber threat detection. Before his professorship he served as a DARPA Program Manager focusing on machine intelligence and image understanding. Dr. Waltzman has also held research positions at the University of Maryland, Teknowledge Corporation, and the Applied Physics Laboratory of the University of Washington.