Dr. Gogate is a Big Data Champion
For three months in 2010 and in 2012, geeky, international competitions were held online and Dr. Vibhav Gogate’s team won in several categories. His team competed against about eight other teams, but this doesn’t mean there wasn’t much competition. Quite the contrary! There are only a handful of teams in the world that have the ability to compete in the probabilistic inference competition organized and supported by the Uncertainty in Artificial Intelligence (UAI) conference. UAI is a top-tier conference in machine learning and artificial intelligence that focuses on probabilistic models for analyzing a large amount of data.
In the UAI competition, each team is asked to write general-purpose software for solving one or more of the three key inference tasks defined over probabilistic models: (1.) Compute the Marginal Probabilities given evidence (e.g., compute the probability of Earthquake given that the Burglar alarm is on), (2.) Calculate the probability of an event (e.g., compute the probability of rain and tornado happening simultaneously), and (3.) Compute the most probable explanation for the evidence (what is the most likely state of the system which explains why the Burglar alarm is on). Before the data is made available for the competition, the organizers take their time to perform the three tasks and calculate the exact answers (probabilities). The teams, however, must write efficient software to perform the tasks in one of three time intervals: (1.) Twenty seconds, (2.) Twenty minutes, or (3.) One hour. Therefore, there are a total of nine categories (3 x 3 = 9) in which a team may compete. For each category, winners are determined by how close the results are to the actual numbers.
For some, all the talk about “Big Data” may be confusing. A definition from Wikipedia states: “Big data is the term for a collection of data sets so large and complex that it becomes difficult to process using on-hand database management tools or traditional data processing applications.” In other words, Big Data is a set of related data and is so big that it is difficult to work with.
Big data analytics can provide valuable insight and help to spot business trends, health trends, social trends, and consumer trends. It can also help to control traffic, prevent diseases, and combat crime. The problem with using big data is that it takes a long time to analyze it. That’s why it is important to develop software techniques to analyze the data quickly and accurately.
For the 2010 and 2012 UAI competitions, Dr.Gogate worked with his PhD and post doctoral advisors, Dr. Rina Dechter from the University of California at Irvine and Dr. Pedro Domingos from the University of Washington. They participated in the following six categories:
- Computing the Marginal Probabilities – within 20 sec., 20 min., and 1 hr.
- Calculate Partition Function Approximations – within 20 sec., 20 min., and 1 hr.
In 2010, Dr.Gogate’s team won four of the six categories in which they participated. In 2012, Dr. Gogate’s team won all six of the categories in which they participated. Clearly, they were able to analyze massive amount of data quickly and accurately! More information about the 2012 UAI competition is available at: http://www.auai.org/uai2012/pascal.shtml.
Future UAI competitions will be organized differently. The same nine categories will be used, but the 3-month window of competition, every several years will be changed to an on-going competition. Starting in 2014, the UAI competitions will be organized like an ongoing “Tennis Ladder” with the team with the most accurate results at the top of the list.
It is an honor that Dr. Gogate has been asked to work with Microsoft Research to help organize the 2014 UAI competition. The analysis of Big Data has a big future and Dr. Gogate is a major player in the field. Currently, four fortunate graduate students at UTDallas, David Smith, Deepak Venugopal, Tahrima Rahman, and Somdeb Sarkhel, are working under his guidance.
Dr. Vibhav Gogate is an Assistant Professor of Computer Science at The University of Texas at Dallas [www.hlt.utdallas.edu/~vgogate]. His current research interests are: Probabilistic Graphical Models, Statistical Relational Learning and Probabilistic Programming. Corporations and businesses, interested in participating in his Big Data research should contact Dr. Gogate via email at: vgogate@hlt.utdallas.edu.
The Department of Computer Science at UT Dallas [www.cs.utdallas.edu] is one of the largest CS departments in the United with more than 750 undergraduate, 500 master, and 125 PhD students. They are committed to exceptional teaching and research in a culture that is as daring as it is supportive.