CS Assistant Professors Aditya Parameswaran and Jian Peng have each received a prestigious CAREER Award from the National Science Foundation (NSF) to support their research. Given by the NSF’s Faculty Early Career Development Program, these awards recognize junior faculty who have the potential to serve as academic role models in research and education.Advancing Open-Ended Crowdsourcing: The Next Frontier in Crowdsourced Data Management
An expert at incorporating humans into data analytics systems, Parameswaran plans to address challenges in managing and optimizing open-ended crowdsourcing. Because there are many tasks that humans do better than computers, especially those that deal with processing images, video, and raw text, crowdsourcing is the main source of high-quality labeled training data for machine learning systems. However, human workers can be expensive, slow, and inaccurate, making these crowdsourcing tasks hard to orchestrate.
Parameswaran’s award-winning dissertation developed optimal ways to leverage crowdsourcing for a variety of fundamental closed-domain data processing tasks. However, open-ended crowdsourcing — where responses from human workers can come from an unbounded set of alternatives — presents a new set of research questions. Despite being widely used, it has not gotten similar attention from researchers, leading to low-quality data and wasted resources. Challenges include deciding which problem-specific open-ended task is appropriate to assign to workers, as well as ascertaining which open-ended responses from workers are correct. Parameswaran’s goal of establishing foundational principles for the efficient data management of open-ended crowdsourcing would allow researchers in machine learning to extend their reach to more challenging domains.
Prior to joining Illinois in 2014, Parameswaran spent a year as a postdoc at MIT CSAIL following his PhD at Stanford University. He has received multiple awards and recognitions, including the IEEE TCDE Early Career Award for contributions to data engineering; outstanding dissertation awards from SIGMOD, SIGKDD, and Stanford University; and five best-of-conference citations for papers at top database and data mining conferences. His previous crowdsourcing research was catalogued in a book that he co-wrote, titled Crowdsourced Data Management: Industry and Academic Perspectives, which also sought to bridge the gap between state-of-the-art research and industrial practice.
Large-scale Biological Network Integration with Applications to Automated Function Annotation
Peng designs efficient algorithms for biological data analysis – software tools that are especially important for extracting meaningful information from genomic sequencing data (which is growing at an exponential rate) and from the large repositories of experimental data now being generated by high-throughput techniques in protein analysis and in biotechnology. A comprehensive understanding of the various functional aspects of a gene or a protein is critical for research in biology and in medicine.
In his CAREER-sponsored research, Peng plans to develop a new computational framework to integrate the large amounts of high-resolution data being generated by those applications, with the goal of enabling the annotation of genome-scale gene functions across species. He also aims to create methods to infer functional homology or analogy between genes from different species.
Peng joined Illinois in 2015 after completing a PhD at the Toyota Technological Institute and then spending a year as a postdoc at MIT. He is a prior recipient of a prestigious Sloan Research Fellowship in 2016, a 2017 PhRMA Foundation Award in Informatics, and a 2016-2017 NCSA Faculty Fellowship. Algorithms developed by Peng and his collaborators have been successful in six scientific challenges, including the Critical Assessment of Protein Structure Prediction (CASP) in 2010 and 2016. Peng’s recent research on mapping a gene’s “social network,” published earlier this year in Cell Systems, could provide insights into fighting human disease.