News & Events
CISE WELCOMES NEW FACULTY MEMBER ALIN DOBRA
October 22, 2003
GAINESVILLE, Fla. --- If new CISE Assistant Professor Alin Dobra has his way, he'll finalize research that develops innovative ways to extract relevant information from large amounts of data in a timely, cost effective, and human-comprehensible way. His long term research plan is to seamlessly integrate approximate query processing, data mining, and relational database technology. Dobra's faculty position encourages him to focus on research while sharing his knowledge teaching undergraduate students three hours per week.
During a two-year internship with Bell Labs while he was pursuing his Doctorate degree at Cornell University, Dobra found out the hard way that freedom for theoretical research in the commercial sector is rare. "Due to the economic situation, there are more and more constraints on research in industry-there's virtually no more pure research being done. And that's across the board, not just at Bell Labs," Dobra said. "In contrast, the academic environment at CISE is one with people who share common interests and there are lots of collaborators." He loves the freedom and encouragement he has now to use statistics and theoretical computer sciences, such as algorithm design, for applications related to databases.
His wife, Delia, and one year old son are also happy to be in Gainesville. "We're crazy about this town," Dobra said, "Especially coming from upstate New York. We're no longer in the middle of nowhere-there are big towns and nice beaches nearby."
When asked what exactly data mining is, Dobra replies, "It's a cool name for a blend of machine learning and databases." He quickly starts sketching out a chart on the large board in his office, which is already covered with what looks like hieroglyphics to the layperson. "I'm a little special in that I do multiple areas," he says as he links machine learning to data mining and databases on his chart.
"One example of my research is addressing the question of how can we do what we want to do for machine learning with very large amounts of data. So we use techniques from databases which were designed to efficiently handle large volumes of data (careful control of disk access, memory management, etc.) to speed up learning algorithms as good as the traditional machine learning algorithms, but hundreds or thousands of times faster."
Dobra explains that in machine learning a core problem that has existed for at least 20 years is classification. This evidences itself in practical situations such as that faced by a company that wants to develop criteria for granting credit. Dobra aims to answer how the data entered on potential customers links to the ultimate decision of whether or not to grant credit to them, taking into account "noise" such as mistakes in data input, criteria that might be poor in predicting repayment, etc.
The idea is to solve situations where customers who were approved for credit didn't behave as projected. By utilizing new discoveries with strong connections between statistics and knowledge linking, Dobra creates computer science applications that can refine decision making, in this example by making the decision to grant credit as informed as possible while eliminating chance.
When not working, Dobra, who came to the United States from Romania, loves to go out driving. One year ago, he learned to drive a manual and he says he likes to keep track of the shifting of gears that go on within the vehicle. "If I could have my dream car, it would be something like a ten year old Porsche," Dobra explains. "I don't want something modern with all the gizmos." When asked if he dreams of being a race car driver, Dobra laughs and says that since he has a wife and young son, it's not an option.
Both he and his wife enjoy reading in their spare time, but they are also both busy with school, Dobra as a professor and his wife as a student. After receiving her Master's in physics from Cornell, Dobra's wife has decided to return to school to get a Bachelor's degree in nursing from UF. Again, Dobra laughs, saying, "She's so intelligent and well-educated I think she's scaring her current professors a little."
CISE is happy to welcome Alin Dobra to CISE and his family to Gainesville.
Writer: Terri Bailey, 352-373-1041, TBailey567@aol.com
Source: Alin Dobra, 352-392-2722, adobra@cise.ufl.edu
