CSE alumnus Daniel Hsu, who earned his Ph.D. in Machine Learning in 2010 from UC San Diego, is now on the computer-science faculty at Columbia University, and he has received a rare honor. Machine learning is a branch of artificial intelligence (AI), and Hsu was just named one of "AI's 10 to Watch" by the IEEE Computer Society publication, Intelligent Systems. Recipients of the honor must have received their Ph.D. degree in the previous five years (between January 2010 and December 2014 for the latest round). The list is compiled every two years to "celebrate 10 young stars in the field" of artificial intelligence. (Photo below of Daniel Hsu by Ryan John Lee)
According to Intelligent Systems (Volume 31, Issue 1, pp. 56-66, January-February 2016), "despite being relatively junior in their career, each one has made impressive research contributions and had an impact in the literature -- and in some cases, in real-world applications as well." Specifically, the publication cited Hsu's "Algorithms for Machine Learning", and his specialization in interactive learning (a subset of machine learning). Interactive learning makes it possible for a computer to learn by hand-labeling a much smaller set of data compared to traditional machine learning (where more hand-labeling is required to recognize future data).
Hsu is a member of Columbia's Data Science Institute, which profiled the algorithmic work that won him the 10 to Watch distinction. In particular, the computer scientist develops algorithms for computer-aided analysis of electrocardiogram to improve the accuracy and speed of heart diagnoses. Hsu's work has also been used in connection with personalized medicine, automated language translation, and methods to reduce noise and preserve privacy while maintaining data integrity on the Internet.
"By shrinking the number of labels needed, the active learning process exponentially speeds up the process of training algorithms to do useful things," according to the Data Science Institute profile, which went on to note that Hsu developed an active learning method while at UC San Diego that was later applied to electrocardiograms that reduced the amount of training data by 90 percent. Hsu's theoretical work also extends to recommendation systems, and he has developed algorithms for Hidden Markov Models (HMMs) such as those used by Siri and Cortana speech recognition to infer written words from a stream of sounds. The work on HMMs has also been applied to Spectacle ENCODE annotations in genomics, notably to infer regulatory changes in the cell from the surrounding chromatin state.
"It is impressive to note that Daniel Hsu's work was already used in Spectacle ENCODE and in EKG analysis," said CSE Chair Rajesh Gupta. "The Columbia University article is well-written and explains the advance, the background in active learning in terms everyone can understand and acknowledges the background training and work here at UC San Diego."
Read the Columbia University Data Science Institute profile of Daniel Hsu.
Learn more about AI's 10 to Watch in IEEE Intelligent Systems