Learning-Aided Peer Prediction
(Yiling Chen, Harvard University)
Abstract:
Information elicitation without verification (IEWV) is a classic problem where a principal wants to design reward rules to elicit high-quality answers of some tasks (e.g. experience of a hotel stay) from strategic agents despite that she cannot evaluate the quality of agents’ contributions. This is also a prevailing problem due to the wide adoption of crowdsourcing. This talk focuses on some of our recent efforts on integrating machine learning techniques into the incentive design for IEWV. In particular, I’ll discuss how we leverage techniques of learning from noisy labels to design peer prediction mechanisms that induce truth telling as a weakly dominant strategy and how we design peer prediction mechanisms that do not always require redundant assignments. Our mechanisms are data driven in the sense that the parameters and some component of the reward rules are learned from the reports of the agents.
The talk is based on joint work with Yang Liu.
Bio:
Yiling Chen is a Gordon McKay Professor of Computer Science at Harvard University and a visiting scholar at Microsoft Research New England. She received her Ph.D. in Information Sciences and Technology from the Pennsylvania State University in December 2015. Prior to working at Harvard, she spent two years at Yahoo! Research in New York City. Her current research focuses on topics in the intersection of computer science and economics. She was a recipient of NSF Career award and and The Penn State Alumni Association Early Career Award, and was selected by IEEE Intelligent Systems as one of "AI's 10 to Watch" in 2011. She is an associate editor for Journal of Artificial Intelligence Research, ACM Transactions on Economics and Computation and ACM Transactions on Social Computing and has co-chaired the 2013 Conference on Web and Internet Economics (WINE’13) and the 2016 ACM Conference on Economics and Computation (EC’16).
Polling the 2020 election
(David Rothschild, Microsoft Research)
Abstract:
Traditional data collection in the multi-billion dollar survey research field utilizes representative samples. It is expensive, slow, inflexible, and its accuracy is unproven; the 2016 election is a crushing blow to its reputation. Intelligence drawn from surveys of non-representative samples, both self-selected respondents and random, but non-representative respondents, is now cheaper, quicker, flexible, and adequately accurate. Further cutting-edge data collection and analytics built around non-representative samples are moving past surveys into gaming, social media, and beyond. I will provide some concrete examples of what went wrong in 2016 and what will go right in 2020.
Bio:
David Rothschild is an economist at Microsoft Research. He has a Ph.D. in applied economics from the Wharton School of Business at the University of Pennsylvania. His primary body of work is on forecasting, and understanding public interest and sentiment. Related work examines how the public absorbs information. He has written extensively, in both the academic and popular press, on polling, prediction markets, social media and online data, and predictions of upcoming events; most of his popular work has focused on understanding the public’s sentiment, an economist take on public policy, and choices in news consumption.
The New New Blockchains and How They Will Transform The World
(Emin Gun Sirer, Cornell University)
Abstract:
This talk will first examine how the first generation of public blockchains, such as Bitcoin and Ethereum, highlighted what is achievable with planetary-scale consensus mechanisms, and discuss why they were a bad fit for business applications. It will then delve into why the second generation, called permissioned blockchains and actively developed by many institutions, are likely to fail for reasons entirely different from the first generation. It will end by outlining a vision for the next generation of blockchain technologies, enabled by recent advances in distributed systems and secure hardware, and discuss their unique features, as well as the novel applications they enable.
Bio:
Prof. Emin Gün Sirer is an associate professor of computer science at Cornell University, and a co-director of the Initiative for Cryptocurrencies and Smart Contracts. His research focuses on distributed systems, databases and large scale services. He played a key role in the development of cryptocurrencies, having developed the first implemented peer-to-peer cryptocurrency with proof-of-work, co-discovered the biggest known fundamental weakness in Bitcoin, and invented new scalable consensus protocols. He has been in the media spotlight as an expert on blockchains, anticipating the DAO hack and helping in its aftermath.
Content Dynamics and "Information Marketplaces"
(Assaf Zeevi, Columbia University)
Abstract:
Content on the Internet is created (and destroyed) at an astounding pace. Traditional curation sites, search engines, Q/A platforms, content recommendation engines and a plethora of other services attempt to provide some structure and help users navigate this vast, and rapidly changing landscape. In this talk we will survey some aspects of content temporal dynamics, primarily focusing on two dimensions that are mostly driven by users. The first focuses on collective attention and the manner in which it contributes to determining the ``shelf life" of content. The second focuses on content that emanates from users, including question-and-answer sites, reviews, ratings and the like.
We will discuss some aspects of the data that are particular to these two strands, some stylized models that pertain, and implications to dynamic learning problems that are central to services and platforms that operate on top of said content. An interesting and central question here, that we will merely touch upon, is quantifying the (temporally changing) value of information and how it potentially influences the various application domains mentioned above. The talk will primarily focus on a few illustrative examples rather than developing any general theory.
Bio:
Assaf Zeevi is the Kravis Professor of Business at the Graduate School of Business, Columbia University. His research focuses on the formulation and analysis of mathematical models of complex systems, with particular research and teaching interests that lie in the intersection of Operations Research, Statistics, Computer Science and Economics. Recent application areas have been motivated by problems in healthcare analytics, dynamic pricing, recommendation engines and personalization, and the valuation and monetization of digital goods. Assaf received his B.Sc. and M.Sc. (Cum Laude) from the Technion, in Israel, and subsequently his Ph.D. from Stanford University. He is the recipient of several research awards including a CAREER Award from the National Science Foundation, an IBM Faculty Award, Google Research Award, as well as several best paper recognitions. Assaf is a member of several editorial boards in his professional community, as well as several advisory boards for companies in the high technology sector.