Two faculty members of the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS) recently received the Faculty Early Career Development (CAREER) Award from the National Science Foundation (NSF) — one of the most prestigious awards for up-and-coming researchers. The award provides funding to support research. 

Yaron Singer, Assistant Professor of Computer Science

Singer will use the grant to develop new algorithms and optimization frameworks to analyze the messy and often unwieldy data from human interactions online.  

Social data — which include everything from purchasing behavior on Amazon to interpersonal interactions on Facebook, Twitter and Reddit — can be too noisy and large to process in a productive way. If that data is going to be used to make predictions and models, researchers need algorithms that can make sense of it.

Singer’s research includes several avenues of exploration, including figuring out how much error and noise in data are tolerable for optimization; establishing a machine learning paradigms for what kind of processes — such as the mechanics of viral content — are statistically learnable; and leveraging inherent structure in data to design more effective algorithms. 

“Social data generates a need to redesign and rethink algorithms and algorithmic frameworks,” Singer said. “If we want to make predications, for example about whether or not content is going to go viral, we need algorithms that can deal with difficult to process data.”

Singer is no stranger to social data. As a graduate student at the University of California, Berkeley, he received the Microsoft Research fellowship and the Facebook fellowship and completed his postdoctoral fellowship at Google Research.

Stratos Idreos, Assistant Professor of Computer Science

Idreos will use his CAREER Award to develop evolutionary data systems, a new class of data systems that can autonomously design and modify the way they store and process data. 

Big data requires efficient storage and processing. As data gets bigger, as applications get more diverse and as new hardware hits the market, data systems have to be continually redesigned to handle data efficiently.  It’s a time-consuming and expensive process. It often takes 5-10 years to design a new system. 

“It is unsustainable,” said Idreos. “So our main question for this research path is how can we make the design process faster and more sustainable in the long term so we can make data systems and the power of analytics accessible to everyone?”

Evolutionary systems, Idreos said, can reduce the amount of work needed to launch or tune a data processing technique by automating large parts of the design process. 

Idreos leads DASlab, the Data Systems Laboratory at SEAS. This year, he was also awarded the TCDE Early Career award by the IEEE Technical Committee on Data Engineering. 

Idreos received his Ph.D. from University of Amsterdam in the Netherlands and was a Scientific Staff Member with the Dutch National Research Center for Mathematics and Computer Science before joining SEAS. For his PhD thesis on Database Cracking he was awarded the ACM SIGMOD Jim Gray Dissertation Award and the ERCIM Cor Baayen award as “Most promising European young researcher in computer science and applied mathematics."