9/8/2020 Laura Schmitt, Illinois CS
Several CS faculty have received new funding from the National Science Foundation during the past few months.
Written by Laura Schmitt, Illinois CS
In the past few months, several Illinois CS faculty have received new NSF research grants that span a range of topics, including data mining, software engineering, autonomous systems, cloud computing, computer architecture, and bioinformatics.
Here’s a summary of each of these newly funded projects.
Knowledge hypercubes - NSF Award #1956151
Jiawei Han and his co-PIs Aidong Zhang (University of Virginia) and Jing Gao (University of Buffalo) received a $1.2 million NSF grant to develop knowledge hypercubes for organizing and retrieving knowledge from biomedical and news event domains. Formed from a massive collection of text documents, these hypercubes aim to allow complex exploration and prediction tasks.
Han and his team are tackling a series of technical challenges, including the design of innovative weakly supervised approaches to organize text documents based on the hypercube structure, the development of novel refinement approaches to automatically verify the information quality within and across cells in knowledge hypercubes by cross-checking within the hypercubes and with external information, and the introduction of an automatic knowledge search pipeline for leveraging knowledge hypercubes for downstream prediction tasks and a hypothesis generation approach for scoring unknown associations between concepts.
Safe and reliable autonomous systems - NSF Award #2008883
Sasa Misailovic and Illinois ECE professor Sayan Mitra (PI) received a $450,000 NSF grant to develop and validate software modules for autonomous systems such as drones and self-driving cars. Their research will focus on several areas of uncertainty, including noisy data from sensors, asynchrony of distributed computation, and heuristic computation of decision-making software.
During the course of the three-year project, they will design a new language for distributed autonomous systems and associated program verification techniques that focuses on uncertainty. The main design goal of the language is to decouple the representation of the program code from both the probabilistic models of uncertainty and underlying solvers.
They will also develop analysis techniques for the language that are based on statistical model checking and probabilistic program analysis.
New technologies to enhance cloud computing - NSF Award #1956007
Josep Torrellas (PI) and Tianyin Xu received an $853,000, three-year grant to enhance cloud computing by addressing challenges around the widespread deployment of modern virtualization technologies, including low-overhead virtual machines known as containers and function-as-a-service environments that enable users to affordably run many small jobs simultaneously.
They will develop a set of novel, transformative computer architecture and operating system technologies that will dramatically improve the security and the scalability of containers and function-as-a-service environments. Specifically, they will explore novel hardware and software designs to filter system calls efficiently and they’ll redesign Translation Lookaside Buffers and page tables that enable thousands of containers to run on the same machine simultaneously.
New software-driven approach to hardware reliability - NSF Award #1956374
A key obstacle for adopting software-driven solutions to hardware errors is that some errors may escape the software stack, leading to unacceptable data corruptions. Sarita Adve (PI), Christopher Fletcher, Darko Marinov, and Sasa Misailovic received a four-year, $1.2 million grant to develop low-cost and practical software-driven approaches to hardware reliability.
Their research will explore new software testing-based techniques to improve the quality and diversity of test inputs used for resiliency analysis; leverage program-analysis and machine-learning methods to make resiliency analysis faster and more accurate for diverse computer architectures; develop formal specifications, optimization strategies, and machine-learning-based methods to harden software using low-cost checkers; and develop techniques to apply resiliency solutions in an incremental and compositional way.
Advancing bioinformatics - NSF Award #2006069
Profile Hidden Markov Models (HMMs) are widely used in bioinformatics, and recently developed "Ensembles of Profile HMMs" (e-HMMs) have been shown to provide improved accuracy in many applications. However, the current designs of e-HMMs are ad hoc, and lack statistical rigor. Tandy Warnow (PI) and Jian Peng received a three-year, $500,000 grant to advance e-HMM method creation.
Specifically, their research will develop rigorous techniques to build e-HMMs, and they will develop methods that use e-HMMs to predict protein structure and function.