Efforts Towards Distributed and Scalable Data Science

Reet Barik, Washington State University
LCF Seminar Graphic

This seminar focusses on my research that has been primarily geared towards scalable data science in the intersection of parallel graph algorithms, combinatorial optimization, AI, and high-performance computing. From the traditional HPC standpoint, the relevant applications include Influence maximization in networks, data summarization, improving diversity in recommendation systems, etc. that can be modeled as a submodular maximization problem. Submodular functions are generally defined over a discrete set with the marginal utility of adding new elements to existing sets exhibiting the property of diminishing returns. This talk will highlight some scalable distributed frameworks to bring down time to solution and increase problem size reach for such applications. It will also cover some recent efforts towards exploring traditional HPC workloads on non-traditional architectures of AI testbeds and evaluation of distributed strategies for LLM training that have the potential of informing research decisions at ALCF going forward.

To add to calendar:

Click on:  https://wordpress.cels.anl.gov/cels-seminars/

Enter your credentials.

Search for your seminar 

Click “Add to calendar”