Su, Sundarrajan Receive CICS 2021 Outstanding Dissertation Awards
Hang Su ‘20PhD and Aditya Sundarrajan ‘20PhD have been announced as the recipients of the 2021 UMass Amherst College of Information and Computer Sciences (CICS) Outstanding Dissertation Awards.
Su’s dissertation, “Deep Neural Networks for 3D Processing and High-Dimensional Filtering,” presents several new types of neural network operations and architectures for data in 3-D and higher-dimensional spaces, and demonstrates how the typical complications with this approach can be mitigated to create a highly efficient, flexible result.
“Any graduate student who introduces a new architectural element that is widely adopted by others can be considered highly successful. Amazingly, Hang Su’s thesis introduces three completely independent and original architectural innovations into deep neural networks,” says Su’s advisor, Professor Erik Learned-Miller. “Each of these contributions is having a large impact, with rapidly growing adoption and citations.”
Su is currently a research scientist in the Learning and Perception Research team at NVIDIA Research. He works in the areas of computer vision, graphics, and machine learning, with a focus on bringing together the strengths of 2-D and 3-D visual information for learning richer and more flexible representations.
Sundarrajan’s dissertation, “Towards Optimized Traffic Provisioning and Adaptive Cache Management for Content Delivery,” presents efficient algorithms to provision user traffic in a large content delivery network (CDN) with the goal of minimizing the cache miss rate, the bandwidth cost of the network, and the energy consumption of the servers.
“Aditya’s thesis provides the scientific foundation for reducing the operating cost and increasing the operational efficiency of content delivery networks,” says Sundarrajan’s advisor, Professor Ramesh Sitaraman. “His thesis is now being used to provision traffic across hundreds of thousands of CDN servers that serve about a quarter of the world’s web and video content. His optimization algorithms can reduce network traffic due to cache misses by as much as 20%—a potential savings of millions of dollars in bandwidth costs. This work is already making an impact on billions of users who access the internet daily.”
Sundarrajan is currently a research scientist at Facebook. He works on building systems to optimize the power consumption of Facebook’s globally distributed datacenter infrastructure.
The CICS Outstanding Dissertation Awards are given annually to graduating doctoral students in recognition of exceptional work and to encourage the highest levels of computer science scholarship and research. Awardees are nominated by CICS faculty.