As artificial intelligence research evolves, new advances and techniques regularly make national headlines.in the School of Computing and Augmented Intelligencepart of Ira A. Fulton School of Engineering At Arizona State University, many faculty members are AI experts and thought leaders expanding the field.
One of these teachers Subbarao Kambangpati, Professor of Computer Science and a global AI thought leader. Kambhampati leads a discussion on generative AI at the first conference. Undergraduate Colloquium of the Faculty of Computing and Augmented IntelligenceIn , we explained the origin, status, and many implications of this rapidly evolving technology. Kambhampati explored tools like ChatGPT, DALL-E and Whisper and their impact on evolving creativity.
The School of Computing and Augmented Intelligence, part of the Ira A. Fulton Schools of Engineering, is a national leader in artificial intelligence, with its undergraduate program ranked 23rd by US News & World Report in 2022. Photo credit: DeepMind on Unsplash
Download full image
Choi Yoo JungAn assistant professor in computer science, he also contributes to the AI field.she is researching Probabilistic modeling reduces the uncertainty of model knowledge to probability distributionAcknowledging the uncertainty of these models helps humans build trust in AI technology.
“Our research presents examples of discriminative patterns, or AI algorithms exhibiting bias,” Choi said. “We show that many of these patterns can exist in probabilistic models, and propose an efficient, accurate, and approximate discriminative pattern miner to find and remove them from probabilistic circuits.”
Her research aims to provide an efficient and straightforward audit of AI models to enforce their fairness and lack of bias. She and her team can suggest better algorithms to remove these discriminatory patterns and create fairer models.
Choi hopes the research will be used to identify and eliminate discriminatory patterns early in the development of probabilistic AI models, allowing researchers to create fairer models from the start. .
“Our school’s talented faculty are constantly working to innovate in the AI field with dynamic research. Ross Massieyevsky, Principal, Professor of Computer Science. “Their passion has positioned our school as a national leader in AI and allowed us to see first-hand important advances in the field.”
Schools also use behavioral language, specifically, mA* it is under development by Citta Balal, Professor of Computer Science. The AI’s action language describes machine commands and instructions and analyzes how to execute requests.
“We are working on developing a foundation for reasoning about actions in multi-agent scenarios, where agents may perform actions not only to achieve an objective, but to deceive other agents. said Baral.
He and his research team are investigating how mA* action languages can bridge functionality in multi-agent domains. This allows him the opportunity to make multiple decisions at once rather than one decision at a time.
The goal of the team developing this language is to take the first step toward creating scalable and efficient automated reasoning and planning systems in the multi-agent domain.
empowering the next generation
In addition to faculty and staff, ASU students are major contributors to AI research. Kaize Ding and Yancheng Wang, computer science graduate students, Yang YingzhenAssistant Professor of Computer Science, and Huang LiuRegents Professor of Computer Science, conducted research on Graph Contrastive Learning (GCL). This is a technique for learning generalizable graph representations by contrasting extended views of input graphs. In computer science, a graph is a group of data points linked together in complex ways.
This technique is used to improve the performance of self-supervised representation learning for graph neural networks (GNNs), a family of deep learning models designed for graph-structured data.
The team is developing a framework — called Simple Neural Networks with Contrastive Structural and Semantic Learning, or S3-CL — addresses the limitations of unsupervised GCL and better captures global knowledge in graphs. The new framework has proven to outperform other unsupervised GCL methods.
Ivan Zvonkova PhD student who will join an assistant professorship in computer science Hannah KernerIn the fall, his lab is also leading research into using machine learning and remote sensing data to create predictive maps of geographic regions. His work with Kerner extends to his next project. NASA Harvestthis mapping will be used to inform indigenous farmers in Maui County, Hawaii, to help combat local food insecurity.
State-of-the-art scientific exchange
One of the forums for sharing innovative research in the AI field is The Japanese Society for Artificial Intelligenceor the AAAI conference fosters discussion among researchers, practitioners, scientists, students and engineers across various AI disciplines.
The 2023 AAAI Conference was held in Washington, DC, with presentations by the aforementioned faculty and students.
Kanbanpati spoke at the conference Bridge: AI and Law There, he discussed the need for “explainability” and transparency in AI technology. In addition, he co-chaired the New Faculty Highlights Program. It spotlights up-and-coming AI professionals early in their careers, such as Choi, who was recognized at the session.
In addition to his involvement, the Kanbangpati student has published four research papers. Representation Learning for Responsible Human Centric AI Workshops and one Artificial Intelligence Workshop for Cybersecurity.
Paulo ShakarianAssociate Professor of Computer Science, worked with Baral on a half-day tutorial sessionThe researchers showcased advances in neural symbolic reasoning (NSR), an emerging field of AI that combines ideas from computational logic and deep learning.
“Some people believe that NSR will be an important part of achieving artificial general intelligence.” Sulu National University and the U.S. Defense Advanced Research Projects Agencyor DARPA.
This tutorial session was intended to educate researchers seeking to understand the current state of NSR research and to engage researchers seeking to apply it to areas such as natural language processing and verification.
Participants explored the framework of NSR, the neurosymbolic approach to deduction, combining NSR with logic and applications, and an overview of the challenges and opportunities facing the field.
“AAAI is one of the premier scientific conferences for AI,” says Shakarian.
