Optimal Learning for Info Collection Course at Princeton
Printer-friendly version
Send to friendCommand, Control, and Interoperability Center for Advanced Data Analysis (CCICADA)
Identifying the source of a radiation release, finding the best path through a congested city to respond to an emergency, learning the best policy for testing cargo containers and evaluating molecular compounds for converting solar rays for portable energy sources are all examples where information must be efficiently collected. All of them are also applications of optimal learning. Researchers at Princeton University, led by Professor Warren Powell, have developed a powerful new method for guiding this process called the knowledge gradient, which makes it possible to quickly sort through many choices with limited budgets all the while taking advantage of the power of correlated beliefs.
Dr. Powell is a member of the Command, Control, and Interoperability Center for Advanced Data Analysis (CCICADA), and this research is the foundation of a new course at Princeton called “Optimal Learning.” The purpose of this research is to develop practical tools for guiding the efficient collection of information in laboratory and field environments. The primary goal is to identify good designs and strategies as quickly as possible in the presence of time and budget constraints.
Two ‐ Day Workshop on Information Analysis Offered
CCICADA and the Statistics Department of Rutgers University will jointly host a two‐day workshop titled “Statistical Issues in Analyzing Information from Diverse Sources ” on the Rutgers campus on May 6‐7, 2010. The workshop will bring together statisticians, applied mathematicians, computer scientists and policy makers to address issues related to combining information. Professor Jim Berger (Duke University) will provide the keynote address. He will be followed by a group of distinguished speakers from statistics, computer sciences and machine learning community. The symposium program and registration details will be posted shortly on the CCICADA website http://ccicada.org/ events.html.