Data-driven problem-solving, 'the modern form of a liberal education'
Ronnie Sircar, chair and professor of operations research and financial engineering (ORFE), speaks about new courses and expanding opportunities to tackle complex problems using data and computation.
This is the fourth in a monthly series of updates from chairs and center directors on news and developments in their departments.
Q. What's new in operations research and financial engineering, and what makes Princeton distinctive?
A. The availability of vast quantities of data and massive computational power has transformed research and practice in our fields, as in many others. This includes machine learning about financial markets and investments, transportation (from traditional freight and taxis to modern ride-sharing apps), health care management, large population dynamics, clinical trials and internet advertising. The core technical disciplines that are the basis of our undergraduate and graduate curricula - namely statistics, probability and optimization - are crucial for many disparate applications and industries.
Our undergraduate program may be considered as the modern form of a liberal education: modern because it is based on science and technology, and liberal in the sense that it provides for broad intellectual development and can lead to many different types of careers. Princeton is also distinctive in not having a business school or a statistics department, which makes us flexible in embracing modern directions in information engineering and applications across industries, without compromising on technical rigor and excellence.
Q. Would you highlight any new endeavors that involve collaborations between ORFE researchers and colleagues in other departments?
A. Professor Amir Ali Ahmadi, in collaboration with Charles Fefferman of mathematics and Clarence Rowley of mechanical and aerospace engineering, received a Multidisciplinary University Research Initiative award from the Department of Defense for a project on verifiable, control-oriented learning on the fly.
The initial phase of this project starts in 2019 and will last for five years. The ultimate goal is to learn the behavior of an unknown, safety-critical dynamical system while controlling it at the same time - for instance, autonomously landing a passenger airplane that has just lost a wing. Its potential impact on the safety and efficiency of our future autonomous systems is enormous.
Q. Are there new developments or plans for undergraduate courses? What kinds of skills can students expect to gain from an ORFE education?
A. We have expanded our popular introductory probability class, ORF 309, to be offered both semesters. This course routinely draws students from other engineering departments, especially computer science, as well as mathematics, economics, physics and others. It has received excellent reviews for its rotating instructors: professors Miklos Racz, Mykhaylo Shkolnikov and Ramon van Handel.
This spring, a new lecturer, Margaret Holen, will introduce a class, "Financial Technology and Data-Driven Innovation," which invites students to consider distinctions between statistical and machine learning models that are decision machines versus discriminative, structural or causal. We also plan to add an undergraduate class in spring 2020 on stochastic networks, including models for understanding and regulating the spread of "fake news," developed by Miklos Racz.
The ORFE education emphasizes the importance of mathematical modelling. Because almost all complex problems include uncertainty, students learn how to develop mathematical models with uncertainty, how to incorporate real-world data into these models, and how to make optimal decisions that improve performance or manage resources effectively. Such a principled and quantitative approach to solving complex problems is of central importance in many different areas of our society.