Top

KAUST master’s degree student wins best poster award at Data Science Summer School

Oct 11, 2018 Students

Samuel Horváth, a KAUST master’s degree student in the statistics and optimization M.S./Ph.D. program, won a best poster award at the Data Science Summer School in Paris in June. File photo.

​-By David Murphy, KAUST News

Samuel Horváth, a master's degree student in the University's statistics and optimization M.S./Ph.D. program, recently won a best poster award at the second edition of the Data Science Summer School (DS3). The event took place in Paris from June 25 to 29, and is one of the largest global data science summer schools. It was co-organized by the Data Science Initiative of École Polytechnique and DATAIA Institute.

The primary goal of DS3 was to provide a series of courses and practical sessions covering the latest advances in the field of data science. Five hundred students, postdoctoral fellows, academics, members of public institutions and professionals from over 30 countries attended the summer school. In a competition featuring 170 posters, Horváth's poster was one of only two to win the best poster award—which included a €500 cash award.

"It was an honor to win the award and a big motivation for future work to win this prize among so many great posters [at the event]," Horváth noted. "I would like to thank KAUST and my supervisor for giving me the opportunity to conduct research and travel to conferences and summer schools where I can present my work—which resulted in this prize."

A focus on empirical risk minimization

Horváth's winning poster entitled "Nonconvex Variance Reduced Optimization with Arbitrary Sampling" is based on a paper of the same name currently under review. The paper features joint work between Horváth and his supervisor Professor Peter Richtárik from the KAUST Visual Computing Center.

Samuel Horváth, a KAUST master’s degree student in the statistics and optimization M.S./Ph.D. program, won a best poster award at the Data Science Summer School in Paris in June. He is pictured here with his poster. File photo.

 ​
"In this work, we focused on empirical risk minimization, which is a big problem in machine learning, as it plays a key role in training supervised learning models, including classification and regression problems, such as support vector machine, logistic regression and deep learning. We provided the first important sampling variants of variance reduced algorithms for empirical risk minimization with non-convex loss functions," Horváth said.

Continued learning

Prior to joining KAUST in 2017, Horváth completed his B.A. in financial mathematics at Comenius University in Bratislava, Slovakia. His research interests at KAUST are at the interface of statistical learning and big data optimization, with a focus on randomized methods for non-convex problems.

In the future, Horváth sees himself continuing to focus on machine learning research, and specifically in optimization for machine learning.

"Machine learning—more generally AI—is a very viable field, and I am excited to see all the new improvements and discoveries," Horváth concluded.

Related stories: