UM Statistical Learning

Rackham Interdisciplinary Workshop

Fall Workshop Schedule

Our workshop meets on Friday at 2:30-4:00pm in the Walker Room on the 5th floor of Haven Hall Please email us to reserve a time-slot if you wish to present work.

Winter Workshop Schedule

  • 2/3 (1:30pm-3:00pm, Walker Room): Ambuj Tewari: A Tutorial on Multi-armed Bandits
    • Abstract: Multi-armed bandit problems (MAB) are very simple sequential decision making problems. At each decision point, the decision maker has to choose from a fixed finite set of available actions. Different actions yield different rewards and the goal is to maximize the total amount of rewards collected. Despite its simplicity, MAB problems allow one to understand the fundamental trade-off between exploration (trying out new actions) and exploitation (repeating actions that have done well in the past). I will talk about the main families of learning algorithms for MAB problems: those that explicitly separate exploration from exploitation (e.g., certainty equivalence with forcing, and epsilon-greedy) and those that don't (e.g., upper confidence bound based, and Thompson sampling aka posterior sampling).
  • 2/10 (3pm-4:30pm, Kuenzel Room, Michign Union): Andrew Gelman: "Theoretical Statistics is the Theory of Applied Statistics: How to Think About What We Do" (joint event with the Foundations of Belief and Decision Making Workshop)
  • 2/17 (2:30pm-3:30pm, Walker Room): Alejandro Pineda: "Stochastic Gradient Descent for Support Vector Machine Learning"
  • 3/10 (2:30pm-4pm, Walker Room): Bill MacMillan: "Automated reference document selection for Wordscores algorithms, with applications to central banking and equities research"
  • 3/17 (2:30pm-3:30pm, Walker Room): Christina Kinane presents research design
  • 3/24 (2:30pm-4pm, Walker Room): Joe Ornstein
  • 3/31 (2:30pm-4pm, Walker Room): Patrick Wu
  • 6/15 (4pm-5:30pm, Eldersveld Room): Michael Littman:: Learning to Act in Multiagent Sequential Environments
    • From routing to online auctions, many decision-making tasks for learning agents are carried out in the presence of other decision makers. I will give a brief overview of results developed in the context of adapting reinforcement-learning algorithms to work effectively in multiagent environments, an area that draws on concepts from both statistical learning theory and game theory. Of particular interest is the idea that even simple scenarios, such as the well-known Prisoner’s dilemma, require agents to work together, bearing some individual risk, to arrive at mutually beneficial outcomes.


Statistical Learning refers to the application and development of statistical frameworks to problems relating to pattern recognition. It is a growing area of research in Statistics and Computer Science. Recognizing patterns in a set of data, either for prediction or inference purposes, lies at the core of scientific pursuits. Not surprisingly, it has influenced the development of scientific investigation in many fields as Economics, Political Science, Finance, Sociology, Biostatistics, Medicine, Marketing, and many other areas.

Some applications of statistical learning involve inference about the flow of information in a network, classifying unseen text data based on previously classified observations (supervised learning and classification), automatically identifying data with similar underlying features (unsupervised learning and clustering), and so on.

Our goal with this interdisciplinary workshop is to provide a forum for researchers and graduate students share how they have applied statistical learning techniques to solve problems in their disciplines. We hope this workshop can provide an environment that facilitates cross-disciplinary information sharing, aiding graduate students, faculty, and researchers in the pursuit of tools to solve problems in their respective disciplines.

Sign up for our email list

Please fill out this form to sign up for our email list. For any questions about the workshop, please contact graduate student coordinators Diogo Ferrari and Blake Miller.

Sign Up


Our workshop will be using Elements of Statistical Learning (ESL) by Trevor Hastie, Robert Tibshirani, and Jerome Friedman and "Machine Learning: a Probabilistic Perspective," by Kevin Patrick Murphy for our review weeks. For those new to statistical learning, an introductory text, Introduction to Statistical Learning (ISL) can be found for free here. In addition, lecture videos corresponding to each chapter in the book can be found here.

Access Digital Book from Umich Library