Introductory Overview Lecture: The Deep Learning Revolution — Invited Special Presentation
JSM Partner Societies
Organizer(s): Ryan Tibshirani, Carnegie Mellon University
Chair(s): Zaid Harchaoui, University of Washington
Deep Learning---broadly speaking, a class of methods based on many-layer neural networks---has witnessed an absolute explosion of interest in Machine Learning in recent years. It has proven to be an extremely useful tool in applications in computer vision, natural language processing, robotics and control, and many other areas. Even apart from these settings, many would argue that Deep Learning is the best "black-box, off-the-shelf" prediction method available.
Should Statisticians now be using Deep Learning for everything? Is this "black-box" really so easy to use, and moreover, can it be opened? Is there room for Statisticians to contribute to the understanding of and/or further development of Deep Learning "models"?
(Spoiler alert on this last question: YES! Come to the IOL to find out more!)
This Introductory Overview Lecture provides a comprehensive overview of some of the most popular/powerful Deep Learning methods, details their application in various data settings, and addresses the questions raised above. Talks will be given by Chris Manning and Ruslan Salakhutdinov, two of the foremost researchers today in Deep Learning. The session will be split into 4 talks of about 25 minutes each, giving by Chris and Ruslan in alternating in fashion.
Introductory Overview Lecture: Examining What and How We Teach at All Levels: Key Ideas to Ensure the Progress and Relevance of Statistics — Invited Special Presentation
JSM Partner Societies
Organizer(s): Rebecca Nugent, Carnegie Mellon University
Chair(s): Mine Cetinkaya-Rundel, Duke University
It has never been a better time to be a statistician. Demand for our profession continues to grow while the emergence of data science has invigorated both industry and academia. However, our education programs are simultaneously facing record numbers of students, the need to keep pace with the rapidly changing set of data-related tools and software development, and unparalleled diversity of career options. Innovation in education and training is a must at all levels, but it can seem daunting - where to begin? This IOL session will highlight changes in the national landscape for introductory level material and both undergraduate and graduate programs in statistics, biostatistics, and data science. We will give an overview of where we are as a field, emphasizing new ideas that could be adopted relatively smoothly, and provoke discussion about where we should be and what it will take to get there.
Introductory Overview Lecture: Leading Data Science: Talent, Strategy, and Impact — Invited Special Presentation
JSM Partner Societies, Caucus for Women in Statistics
Organizer(s): Ming Li, Amazon
Chair(s): Martha Gardner, GE
During the past decade, vast amount of data has become available and readily accessible. To analyze these huge amount of data, big data infrastructure, algorithms, and methodologies have been developed and matured, which powers effective data science projects in tech companies and traditional industrial sectors. In this lecture, data science leaders will demystify real-world data science applications at scale and illustrate how to become an excellent data scientist, how to build a high-impact data science team, how to design data science curriculum and how to lead with statistics.
Introductory Overview Lecture: Multivariate Data Modeling with Copulas — Invited Special Presentation
JSM Partner Societies, Caucus for Women in Statistics
Organizer(s): Christian Léger, Université de Montréal
Chair(s): Bruno Rémillard, HEC Montreal
Regarded as an esoteric concept 30 years ago, copulas and copula models feature today among the most powerful and appealing ways of accounting for dependence in multivariate data. This modeling strategy has found numerous applications in finance, insurance, biostatistics and environmental sciences. This introductory overview lecture, delivered by two active researchers in the area, will describe in simple terms the fundamental principles of this approach and provide concrete illustrations of its use.
In the first part, the notions of copula and copula models will be introduced. The fundamental role of rank-based techniques for inference purposes will be highlighted. Various tools for model construction, fitting and validation will first be presented in the vanilla case of multivariate continuous data without covariates. It will then be seen how extensions can lead, e.g., to powerful tests of independence for sparse contingency tables and new effective ways of analyzing multivariate time series data.
In the second part, selected examples of advanced copula modeling will be discussed. Ways of combining GLMs with copulas will be illustrated. Various strategies for copula modeling of high-dimensional data will also be sketched, with special emphasis on hierarchical dependence structures and sparsity. Finally, techniques for infering a hierarchical dependence structure from data will be outlined. Illustrations will be drawn from the fields of insurance, finance, and hydrology.
Late-Breaking Session: Addressing Sexual Misconduct in the Statistics Community — Invited Special Presentation
JSM Partner Societies, Caucus for Women in Statistics, Committee on Women in Statistics
Organizer(s): Stephanie Hicks, ASA Committee on Women in Statistics
Chair(s): Keegan Korthauer, Dana-Farber Cancer Institute
2:05 PM
Addressing Sexual Misconduct in the Statistics Community Leslie McClure, Drexel University; Kristian Lum, Human Rights Data Analysis Group; Kerrie Mengersen, Queensland University of Technology; Lance Waller, Emory University; Dianne Cook, Monash University; Emma Benn, Icahn School of Medicine at Mount Sinai; Brian Millen, Eli Lilly
Introductory Overview Lecture: Reproducibility, Efficient Workflows, and Rich Environments — Invited Special Presentation
JSM Partner Societies
Organizer(s): Ryan Tibshirani, Carnegie Mellon University
Chair(s): Jacob Bien, University of Southern California
With computing playing an increasingly central role in statistical research, the proliferation of tools, environments, and languages has increased both the power and the complexity of modern data analysis.
When encountering the results of an analysis, several questions arise:
Are the computations accurate? What results were actually computed? Is the computation robust to changes in the size or structure of the data? How were the model parameters tuned? How do the results change when the parameters are adjusted? Are the results generalizable? Are they reproducible?
This session looks at ways to answer these question, exploring a range of issues surrounding the effective use of computing in statistical research and data analysis. The focus is in particular on getting the most out of rich environments, building efficient workflows, and organizing computations to encourage validity, reproducibility, and collaborative sharing. The session will begin with a framing of the issues and overview of current methods by Victoria Stodden, and then delve into more specific issues in talks by Christopher Genovese and Hadley Wickham.
Introductory Overview Lecture: The Statistical and Data Revolution in the Social Sciences — Invited Special Presentation
JSM Partner Societies, Social Statistics Section
Organizer(s): Adrian Raftery, University of Washington
Chair(s): Adrian Dobra, University of Washington
The past decade has seen rapid progress in the development and application of statistical methods for the social sciences, spurred by a huge expansion in available social science data, new kinds of social science data bringing novel statistical challenges, and the establishment of several interdisciplinary centers and institutes in US universities on this interface. This lecture will provide an introductory overview highlighting new statistical methods for three areas of social science where the impact of statistics has been expanding rapidly: demography, social network analysis, and criminology.
