10 am - 12 pm EST
The tutorial will show how Bayesian Optimization (BO) methods can be leveraged by computer systems architects. BO is a powerful approach for solving many challenging problems, but has received limited applicability due to limited expertise in our community and higher barrier to entry. With this tutorial, we are trying to change that. Our group has released multiple BO-based open-source infrastructure and published papers to demonstrate how BO methods can be applied to many of the problems our community faces regularly.
Rohan works on resource management and optimizations in large scale HPC, cloud computing and serverless computing systems.
Tirthak works on quantum computing and HPC. He is interested in exploring trade-offs among fidelity, performance, and efficiency.
Tiwari’s research focuses on innovating new methods to improve the efficiency, scalability, and cost-effectiveness of large-scale parallel computing systems.
10 - 10:55 am
Here we will discuss the basics of BO, how it performs optimization, and for which kind of problems it will be useful. It will also show how to perform various optimizations over traditional BO models for specific applications. We will discuss the uses cases of BO in different areas of computer systems like data-center management and scheduling.
11:05 am - 12 pm
In this session we will discuss the uses cases of BO in different areas of computer systems like data-center management, cloud resource partitioning, auto-tuning, scheduling, etc. Then we will provide a hands on coding tutorial in Python on both custom implementation of a BO, as well as its implementation using standard libraries.
Develop a free site - Learn more