A School of Computer Science team won a 2022 AI4AI Meta Research Award for their work to reduce the cost of artificial intelligence techniques that better automate and optimize the computational performance of machine learning systems.
Zhihao Jia, an assistant professor in the Computer Science Department, earned the nearly $50,000 award with Ph.D. students Yue Zhao and Zhihao Zhang for their proposal, "An AI-Assisted System for Sustainable and Affordable AI."
Existing machine learning systems have surpassed human predictive performance in many real-world tasks such as game play and image generation, but they still rely heavily on human effort to manually optimize deploying computational models on specific hardware platforms. This manual optimization takes time to design and implement. The team aims to change that.
"The computational requirement of AI-based techniques has been quickly growing over the past decade, resulting in high carbon footprint and monetary cost to deploy them in our daily lives," said Jia, the project's principal investigator. "Our goal is to leverage AI techniques to make them more affordable, energy-efficient and inclusive, allowing people to benefit from this modern technology at a much lower financial and environmental cost."
The team's project was one of nine selected from 30 proposals. For more information about the award, visit Meta's website.