RAIN schedule for Autumn Quarter 2014-15
The RAIN seminar is held on Wednesdays from 12:00-1:00pm in Y2E2 101. And yes, lunch is provided!
|January 21||Edith Cohen||Scalable mining of massive networks: Distance-based Centrality, Similarity, and Influence|
|February 11||Glen Weyl||Quadratic Voting|
|February 18||Paul Milgrom||Auctions and Decentralization in Complex Networks|
|March 11||Anindya Ghose||Randomized Field Experiments in Mobile Marketing|
Google Calendar for RAIN
Previous year's talks
Archived talks can be accessed here.
Talk AbstractsScalable mining of massive networks: Distance-based Centrality, Similarity, and Influence
Edith Cohen, Tel-Aviv University
Graphs are ubiquitous in representing interactions between entities: communication, social relations, clicks, likes, follows, views, hyperlinks, transactions, and more. Structural properties of a corresponding graph successfully model fundamental properties of entities and their relations with applications to ranking, recommendations, attribute completion, and clustering.
The shortest-path and reachability relation are the basis of natural measures of centrality, influence, and similarity. The measures can be defined over a static set of edges or, for robustness, over a distribution derived from the set of interactions.
In the talk, I will first present and motivate a unified view of distance-based measures. I will then overview algorithmic and estimation tools from my own work which support scalable computation of these measures on massive networks. The approach is based on computing sample-based sketches which capture the relation of each node to all others. The sketches, with appropriate carefully derived estimators, are then used to provide quick high-confidence approximate answers for similarity, centrality, and influence queries.
An advantages of distance-based measures over others based on the paths ensemble (random-walk/page-rank, Katz, resistance) is the combination of high scalability and application to asymmetric (directed) interactions, which are often harder to work with.
Some of the material is based on joint work performed at the MSR Silicon Valley Lab, with Daniel Delling, Andrew Goldberg, Moises Goldzmidt, Fabian Fuchs, Thomas Pajor, and Renato Werneck.
Bio: Edith Cohen is (visiting) full professor at Tel Aviv University. Until 2014 she was a Principal Researcher at Microsoft Research (Silicon Valley) and before that between 1991 and 2012 she was at AT&T Labs (initially AT&T Bell Laboratories). She was a visiting professor at UC Berkeley in 1997. She received a Ph.D in Computer Science from Stanford University in 1991. Her research interests include algorithms, mining and analysis of massive data, optimization, and computer networking. She is a winner of the IEEE ComSoc 1997 Bennett prize, and an author of 20+ patents and 100+ publications.
Glen Weyl, University of Chicago/MSR
While the one-person-one-vote rule often leads to the tyranny of the majority, alternatives proposed by economists have been complex and fragile. By contrast, we argue that a simple mechanism, Quadratic Voting (QV), is robustly very efficient. Voters making a binary decision purchase votes from a clearinghouse paying the square of the number of votes purchased. If individuals take the chance of a marginal vote being pivotal as given, like a market price, QV is the unique pricing rule that is always efficient. In an independent private values environment, any type-symmetric Bayes-Nash equilibrium converges towards this efficient limiting outcome as the population grows large, with inefficiency decaying as 1/N. We use approximate calculations, which match our theorems in this case, to illustrate the robustness of QV, in contrast to existing mechanisms. We discuss applications in both (near-term) commercial and (long-term) social contexts.
Bio: Glen Weyl is a Researcher at Microsoft Research New England, an Assistant Professor of Economics and Law at the University of Chicago and an Alfred P. Sloan Research Fellow. He was Valedictorian of Princeton University as an undergraduate in 2007, received his PhD also from Princeton in 2008 and then spent three years as a Junior Fellow at the Harvard Society of Fellows before joining the faculty at Chicago. His research focuses on pure and applied price theory, especially applications to industrial and tax policy, as well as the intersection between economics and related fields such as law and philosophy. Outside his academic work, Glen co-founded a start-up commercializing Quadratic Voting and consults for platform start-ups through Applico Inc.
Auctions and Decentralization in Complex Networks
Paul Milgrom, Stanford
Economists' analyses are most often based on simplifying assumptions that enable the analyst to focus on the roles of prices, substitution, and decentralized decision making. Computer scientists most often take a sharply different approach, focusing on worst cases and general algorithmic treatments, in which simple economic structures are mostly hidden from view. We combine elements from these two perspectives in an attempt to develop solutions for high-stakes auction market design problems, especially the reallocation of radio spectrum in North America.
Bio: Paul Milgrom is the Shirley and Leonard Ely professor of Humanities and Sciences in the Department of Economics at Stanford University and professor, by courtesy, at the Stanford Graduate School of Business. He is a member of both the National Academy of Sciences and the American Academy of Arts and Sciences and a winner of the 2008 Nemmers Prize in Economics and the 2012 BBVA Frontiers of Knowledge award.
He is best known for his contributions to the microeconomic theory, his pioneering innovations in the practical design of multi-item auctions, and the extraordinary successes of his students and academic advisees. According to his BBVA Award citation: “Paul Milgrom has made seminal contributions to an unusually wide range of fields of economics including auctions, market design, contracts and incentives, industrial economics, economics of organizations, finance, and game theory.” According to a count by Google Scholar, Milgrom’s books and articles have received more than 60,000 citations.
Randomized Field Experiments in Mobile Marketing
Anindya Ghose, NYU Stern
The explosive growth of smartphones and location-based services (LBS) has contributed to the rise of mobile advertising. In this talk, we present results from multiple randomized field experiments that are designed to measure the effectiveness of mobile marketing and promotions. In the first experiment, using data from a location based app for smartphones, we measure the effectiveness of mobile coupons. The aim is to analyze the causal impact of geographical distance between a user and retail store, the display rank, and coupon discounts on consumers’ response to mobile coupons. In a second large scale field study we examine the role of contextual crowdedness on the redemption rates of mobile coupons. We find that people become increasingly engaged with their mobile devices as trains get more crowded, and in turn become more likely to respond to targeted mobile messages. The change in behavior can be accounted for by the phenomenon of “mobile immersion”: to psychologically cope with the loss of personal space in crowded trains and to avoid accidental gazes, commuters can escape into their personal mobile space. In turn, they become more involved with targeted mobile messages they receive, and, consequently, are more likely to make a purchase in crowded trains. These studies causally show that mobile advertisements based on real-time static geographical location and contextual information can significantly increase consumers’ likelihood of redeeming a geo-targeted mobile coupon. However, beyond the location and contextual information, the overall mobile trajectory of each individual consumer can provide even richer information about consumer preferences. In a third study, we propose a new mobile advertising strategy that leverages full information on consumers’ offline moving trajectories. To examine the effectiveness of this new mobile trajectory-based advertising strategy, we designed a large-scale randomized field experiment in one of the largest shopping malls in the world. We find that mobile trajectory-based advertising can lead to highest redemption probability, fastest redemption behavior, and highest satisfaction rate from customers at the focal advertising store. Various practical implications for mobile marketing are discussed.
Bio: Anindya Ghose is a Professor of IT and a Professor of Marketing at New York University's Leonard N. Stern School of Business. He is the co-Director of the Center for Business Analytics at NYU Stern. He is the Robert L.& Dale Atkins Rosen Faculty Fellow and a Daniel P. Paduano Fellow of Business Ethics at NYU Stern. He has been a Visiting Associate Professor at the Wharton School of Business. He also serves as the Scientific Advisor to 3TI China. He was recently named by Business Week as one of the "Top 40 Professors Under 40 Worldwide" and by Analytics Week as one the "Top 200 Thought Leaders in Big Data and Business Analytics". His research has been recognized by 10 best paper awards and nominations. He has published more than 75 papers in premier scientific journals and peer reviewed conferences, and has given more than 200 talks internationally. His research has been profiled and he has been interviewed numerous times in the BBC, Bloomberg TV, CNBC, China Daily, Financial Times, Fox News, Forbes, Knowledge@Wharton, Korean Broadcasting News Company, Los Angeles Times, MSNBC, NBC, New York Times, New York Daily, National Public Radio, NHK Japan Broadcasting, Reuters, Time Magazine, Washington Post, Wall Street Journal, Xinhua, and elsewhere.