As research in the field of artificial intelligence evolves, new advances and technologies regularly make national headlines. At the School of Computing and Augmented Intelligence, part of the Ira A. Fulton Schools of Engineering at Arizona State University, many faculty members are among the AI experts and thought leaders advancing this field.
“Our school’s exceptional faculty members continually strive to innovate in the field of AI with dynamic research,” said Ross Maciejewski, director of the School of Computing and Augmented Intelligence and professor of computer science. “Their passion has positioned our school as a national leader in AI and allows us to see critical advances in the field firsthand.”
YooJung Choi, assistant professor of computer science, is one of these researchers. Her work focuses on probabilistic modeling, an essential component of AI that examines the uncertainty in the models’ knowledge by explicitly representing it as a probability distribution. Acknowledging the uncertainty in these models helps people build confidence in AI technologies.
“For our research, we introduce patterns of discrimination, or examples of when AI algorithms show bias,” says Choi. “We show that a large number of these patterns can exist in a probabilistic model, and then propose efficient, exact, and approximate discrimination pattern miners to find and remove them from probability circles.”
Her research aims to provide an efficient and easy-to-understand test of AI models to reinforce their fairness or impartiality. She and her team are then able to propose better algorithms for removing these discrimination patterns to create fairer models.
Choi hopes this research will be used to identify and eliminate patterns of discrimination early in the development of probabilistic AI models, allowing researchers to create fairer models from the start.
The School of Computing and Augmented Intelligence is also researching action language, specifically a new language called mA* being developed by Chitta Baral, a professor of computer science. Action languages in AI describe commands and instructions for machines and analyze how they can carry out requests.
“We are working to develop a basis for reasoning about actions in a multi-agent scenario, where an agent can perform actions not only to achieve a goal but also to deceive other agents,” says Baral.
He and his research team are investigating how their mA* action language can bridge the capabilities of a multi-agent domain that allows multiple choices at once rather than a single choice at once.
The team’s goal in developing this language is to take a first step towards creating scalable and efficient automated reasoning and planning systems in multi-agent domains.
empowering the next generation
In addition to faculty, ASU students also make important contributions to leading-edge AI research. Computer science students Kaize Ding and Yancheng Wang work closely with Yingzhen Yang, an assistant professor of computer science, and Huan Liu, a regents professor of computer science, to conduct research on contrastive graph learning, or GCL, a generalizable learning technique using graph representations Contrasting the expanded views of the input graph. In computer science, a graph is a group of data points that are connected in complex ways
This technique is used to improve the performance of self-supervised representational learning of graphical neural networks or GNNs, which are a family of deep learning models developed for graph-structured data.
The team is developing a framework called Simple Neural Networks with Structural and Semantic Contrastive Learning, or S3-CL, to address the limitations in the unsupervised GCL, making it easier to capture global knowledge in a graph. The new framework has proven that it can outperform other unsupervised GCL methods.
Ivan Zvonkov, a prospective graduate student who will join computer science assistant professor Hannah Kerner’s lab in the fall, is also leading the research using machine learning and remote sensing data to create predicted maps of geographic regions. His work with Kerner also extends to a project with NASA Harvest using this mapping to inform Indigenous farmers in Maui County, Hawaii to help address local food insecurity.
Leading scientific exchange
One of the forums for the exchange of innovative research in the field of AI is the Association for the Advancement of Artificial Intelligence (AAAI) conference, which promotes exchanges between researchers, practitioners, scientists, students and engineers from a range of AI disciplines.
The 2023 AAAI Conference was held in Washington, DC and included presentations from all of the above faculty members and students presenting research from the School of Computing and Augmented Intelligence.
Subbarao Kambhampati, Professor of Computer Science and global AI thought leader, spoke as part of the conference’s Bridge: AI and Law program. There he discussed the need for “explainability” and transparency in AI technologies.
In addition, Kambhampati co-chaired the New Faculty Highlights program, which spotlights promising AI professionals early in their careers, such as: B. Choi, who was awarded at the meeting.
In addition to his participation, Kambhampati’s students presented four research papers at the Representation Learning for Responsible Human-Centric AI workshop and one at the Artificial Intelligence for Cyber Security workshop.
Paulo Shakarian, Associate Professor of Computer Science, collaborated with Baral to create a half-day tutorial session. Researchers presented advances in neuro-symbolic reasoning, or NSR, an emerging field of AI that combines ideas from computational logic and deep learning.
“Some people think that NSR will be an important part of artificial general intelligence,” says Shakarian, who moderated the mini-course with colleagues from Argentina’s Universidad Nacional del Sur and the US Defense Advanced Research Projects Agency (DARPA), in addition to Barall.
The tutorial session aimed to train researchers who wish to understand the current landscape of NSR research and to attract those who wish to apply NSR research in areas such as natural language processing and verification.
Participants explored an overview of the framework of NSR, neuro-symbolic approaches to reasoning, combining NSR with logic and applications, challenges and opportunities faced by this field.
“AAAI is one of the best, if not the best, academic conference in the field of AI,” says Shakarian, “so it was a great honor to hold a session there to present our tutorial.”
