Each year, 55 to 60 young researchers win the CRII (Computer and Information Science and Engineering Research Initiation Initiative) award from the National Science Foundation (NSF). In the past two years, three Principal Investigators from USC’s Information Science Institute (ISI) have received this award. While the more widely known NSF CAREER award applies to tenure-track faculty, the CRII program clearly enables early-career research faculty to launch their research career.
Filip Ilievski, a research scientist at ISI’s Center on Knowledge Graphs, received the CRII award on March 15 this year. In 2021, Muhao Chen from the Artificial Intelligence Division and Loïc Pottier from the Computational Systems and Technologies Division both scored one. This program aims to provide essential resources and funding enable promising early-career researchers to launch their research careers, with sufficient funds for four years of graduate student support.
By laying the foundation for funding their research goals, the NSF CRII Award enables ISI Principal Investigators Ilievski, Chen, and Pottier to hire doctoral students and grow their team to pursue their own research programs. For them, it’s about integrating common sense insights into AI technologies, automatically acquiring structured data from unstructured text., and unite supercomputers and distributed systems to optimize machine learning algorithms, respectively.
Yolanda Gil, ISI’s director for major strategic initiatives in AI and data science and research professor of computer science, spoke about the importance of the NSF program. “The CAREER and CRII programs have been primary vehicles for nurturing the research and education efforts of early career researchers,” Gil said. “Given the incredible talent of young ISI recruits in full-time research positions, the CRII program is a perfect fit.
Gil has also served as a reviewer on committees that evaluate proposals for the NSF CRII program. “I was deeply impressed with the quality of the proposals submitted and the very competitive selection process,” said Gil.
Ewa Deelman, a computer science research professor at ISI, mentored Pottier throughout his research efforts. “I am very happy for Loïc to have received the CRII prize, it is a very good first step for him in the development of his own research program centered around his HPC [high performance computing] interests.”
Pedro Szekely, director of ISI’s Artificial Intelligence Division and associate research professor of computer science, guided Ilievski and Chen to varying degrees through the proposal process. “The work they do isn’t progressive, it’s innovative,” Szekely said. Successful applicants demonstrate that their idea is novel while presenting their plans for success in the submitted proposal.
Sometimes the first time is the charm. This was the case of Filip Ilievski, a researcher at the ISI. After submitting its first-ever proposal, it became the latest ISI winner of the NSF CRII program, a leader in creating AI agents that provide common-sense explanations of open-world narratives.
Inspired by AI’s deficits in using common sense knowledge when interacting with people, Ilievski designed his research proposal with the eventual goal of improving AI assistants for older adults with dementia. dementia and children with autism spectrum disorders (ASD).
His research proposal emphasizes the importance of integrating common sense into AI technologies.
Ilievski referred to the concept of “substitution,” an integral facet of common sense AI technology. “When the stories are different, but they are based on the same axiom, it’s called substitution.” For example, if a child is cold but there is no jacket available, AI technology can help them understand that another available object, such as a blanket, can perform the same function. Ilievski cited this example as an illustration of how this technology can potentially be used to help children with ASD.
Based on knowledge statements written by experts, the machine is then able to see and learn from those statements to fill in the gaps for future similar situations. “If you are able to teach a computer a certain kind of story, then it would be able to understand different stories,” Ilievski said. “And once that’s done, it could help human users do the same over time.”
For Ilievski, this project is only the beginning. He hopes to continue to expand his work on common sense technology to achieve the ultimate goal of providing meaningful social assistance to vulnerable populations and improving the communication skills of AI systems.
Muhao Chen, research director at ISI and research assistant professor in computer science, uses this prize to complete his study on the automatic acquisition of structured information from unstructured texts. In AI terms, this structure is captured in knowledge graphs (KGs), which provide both open-world and domain-specific representations of knowledge that are integral to many AI systems. It also seeks to make structured knowledge transferable across different scientific languages and fields, such as biology, genomics and proteomics.
“My long-term goal is to help the machine understand nature. If he understands nature, he understands what we say, what we know about the world, and how we communicate with machines,” Chen said.
The study proposes a framework that seeks to systematically improve the robustness of learning and inference for data-driven knowledge acquisition models.
And its real-world applications are vast, like helping doctors understand disease targets. “We also apply this kind of technology to disease prediction,” Chen said. “It’s a way to help doctors observe patients. By making predictions, it promotes precision drug design to target disease, an important area of medical research.
Chen also plans to expand automatic knowledge acquisition for software-related online forums, with the ultimate goal of reducing the cognitive effort of software developers.
Loïc Pottier, researcher at ISI, received the NSF CRII award for his research proposal on the advancement of supercomputing and cloud computing to develop machine learning.
But supercomputing isn’t a concept most people are familiar with, so when Pottier explained it, he devised an analogy to help visualize its capabilities. “Think about your laptop and multiply your laptop by a hundred thousand,” Pottier said. “You interconnect all these laptops [to process data together] and you have a supercomputer. Supercomputers are largely found in government facilities like the US Department of Energy to perform scientific calculations, such as climate change simulations or weather forecasts.
Pottier intends to combine supercomputing and machine learning to optimize algorithms. Machine learning (ML) algorithms have become key elements in many scientific fields over the past few years, and they exist virtually everywhere. When datasets get larger, machine learning requires extensive computational capabilities, and that’s where Pottier’s research comes in.
“I’m taking a machine-learning algorithm, running it faster on high-performance computing machines, and spending less energy and fewer resources,” Pottier said.
“If I can help them speed up machine learning algorithms, it will help them run larger, more accurate simulations,” Pottier continued. By improving and optimizing the base layer of data processing, major software advantages emerge that can accelerate research, such as our understanding of COVID-19 simulations or climate modeling.
Posted on April 11, 2022
Last updated on April 11, 2022