IEORE8100 Advanced Topics In IEOR

1-3 pts. faculty adviser's permission. Selected topics of current research interest. May be taken more than once for credit. Please refer to the course site for more information.

Spring 2018 Topic Courses:

IEOR E8100-001 Reading Group on Behavior Finance (3pts)
Instructor: Prof. Xunyu Zhou
This PhD reading course will center around behavioral economics and finance, but will also involve relevant research topics on time-inconsistent dynamic optimization and machine learning. Students will take turns to present papers, either important ones in literature or their own research papers, and discuss about possible open research problems. The objective is to stimulate interest in the related fields, to motivate new problems, and to inspire innovative approaches to solve those problems. 

IEOR E8100-002 Mean Field Games and Interacting Particle Systems (3pts)
Instructor: Prof. Daniel Lacker
PhD course on large-population game-theoretic models and their related mean field (i.e., infinite-population) limits. Topics include: weak convergence of probability measures & Wasserstein metrics, static games, dynamic (continuous-time) interacting diffusions and the McKean-Vlasov limit, stochastic differential games, mean field games, the master equation. The emphasis is on theory, but many examples from recent research will be covered in detail. Will assume a working knowledge of probability and stochastic calculus, and some experience with the basics of stochastic optimal control theory would help but is not strictly necessary, as it will be reviewed briefly.

IEOR E8100-003 Stochastic Inventory Theory (3pts)
Instructor: Prof. Martin Reiman
The ongoing need for a fundamental understanding of inventory management supports the old adage that ‘the more things change, the more they stay the same’. Inventory theory was introduced in the 1950s by such luminaries as Arrow, Bellman, Karlin, and Scarf (among others), motivated by problems in the industrial economy of that time, and has been an important part of OR ever since. Although there have been many dramatic changes in technology and trade since then, leading to such things as ecommerce, offshoring and the sharing economy, the fundamental issue remains the same: assuring adequate supply to meet demand in a manner that reduces both shortage and waste (in order to minimizes costs). The purpose of this class is to introduce students to some of the classical models of stochastic inventory theory along with the techniques used to solve them, and work done over the decades since, with an emphasis on recent work motivated by current concerns.

IEOR E8100-004 Networks: Games, Contagion, and Control (3pts)
Instructor: Prof. Agostino Capponi
Networks are ubiquitous in our modern society. Economic and social networks have been used extensively to model a variety of situations, in which individual decision-makers are affected by the choices of their peers in the network. For instance, the choices of individuals regarding which products to buy or whom to vote for are usually influenced by their friends and colleagues. The decision of an individual or a firm on whether or not to adopt a new technology (new software, messaging service, etc.) depends on who among their social or professional network are adopting that technology as well.

Banks in financial networks may coordinate private or subsidized bailouts and rescue insolvent banks so as to stop financial contagion. The decision of an individual to become a criminal depends heavily on the behavior of others in his/her social network: more connections to criminals yield a higher profitability in the crime business and thus a higher chance of engaging in criminal activities. This course will introduce the main mathematical models for the study of these networks. It will discuss game theoretical and dynamic optimization techniques, which can be used to analyze a wide variety of these networks, including their resilience to shocks, the amplification effects resulting from their topological structure, and how the strategic behavior of network agents shapes the performance of the network. For more information, please review the course syllabus.

IEOR E8100-006 Reinforcement Learning (3pts)
Instructor: Prof. Shipra Agrawal
Reinforcement learning is a powerful paradigm for learning and sequential decision making, and it is relevant to an enormous range of tasks, including robotics, game playing, consumer modeling and healthcare. The course aims to provide a hands-on introduction to the state-of-the-art techniques in reinforcement learning.

We will begin with foundations of sequential decision making and Markov Decision Processes, and quickly move on to the core challenges and recent approaches for large-scale reinforcement learning. Through a combination of lectures, assignments, readings and project, the students will become well versed in the basic RL algorithmic techniques as well as deep RL based algorithmic techniques. The assignments will involve implementing these techniques to solve OpenAI gym environments using Python (Tensorflow + numpy). We will read recently published articles on RL and utilize those ideas to potentially improve our implementations.

IEOR E8100-005 Great Presentations (Required for 2nd year doctoral IEOR students)
Instructor: Janet Kayfetz
Great Presentations is a hands-on, practice-intensive course that focuses on the most important elements of excellent formal presentations. The fundamental principles discussed and practiced in class can be applied in a variety of contexts, including the short research or lab talk, the formal conference presentation, a poster presentation, a job talk, an interview, a class presentation.

500 W. 120th St., Mudd 315, New York, NY 10027    212-854-2942                 
©2014 Columbia University