Priya Panda Receives Google’s Research Scholar Program Award

Priyadarshini Panda, assistant professor of electrical engineering, is the recipient of an award from Google's Research Scholar Program.

Created in 2021, Google's Research Scholar Program supports early-career faculty working on cutting-edge research across many areas in computer science, including machine perception, quantum computing, and other related fields.

Today, owing to IoT, copious amounts of data are accumulated in millions of edge devices, such as mobile phones, traffic cameras, and trackers. However, with increasing concerns around data privacy, practical AI applications for IoT are shifting towards privacy preserving distributed learning. However, distributed learning with SNNs remains an unexplored area, with most of the previous research limited to classic image classification tasks with models trained on a single device. Despite being potentially energy efficient, SNNs are not widely adopted in practical applications.

Panda's project, "Private Explainable & Robust Distributed Learning with Spiking Neural Networks (SNNs)," will explore methods to enhance the end-to-end capability of training SNNs geared towards applications beyond static image recognition in a privacy preserving distributed learning paradigm. As SNNs will find usage in real-world applications such as medical robots, drones, "explainability" – the "why" behind a prediction made by an AI model – in addition to performance is critical. With this award, Panda will further explore how to enable low-power explainable intelligence with SNNs for edge computing platforms.

"Spiking Neural Networks is an emerging area that offers us a pathway to approach brain-like efficiency," Panda said. "The award will help further boost interest in the area, and hopefully foster more collaborations across different disciplines, from engineering, computer science as well as neuroscience to fill the gaps in SNN research."

The award comes with $60,000 in unrestricted funds that will primarily be used for student research training and conducting workshops on spike-based computing to discuss this research direction with others in the community.