AWS/Caltech ML and HPC Workshop #1: Seismology and Time Domain Astronomy

AWS/Caltech Machine Learning and High Performance Computing Workshop #1: Seismology and Time Domain Astronomy

We are excited to announce a series of joint Caltech/Amazon technical workshops as part of our continuing partnership with AWS. This series will explore how AWS' technologies and services can be deployed to enhance scientific discovery. The initial focus will be on applications in seismology and time domain astronomy. However, each session will include a general introduction and an opportunity to work on your own problems.

Each 2 hour workshop will use the following format:

• Presentation by AWS covering their technologies, services, and a field-specific example (1 hour)
• Hands-on session where attendees can work on their own use cases. AWS technical staff will be there to offer assistance (45 minutes)
• Discussion (15 minutes)

HPC - Time Domain Astronomy – Session Summary:

In this workshop, we will discuss multiple methods for performing large-scale scientific computations in the cloud and explore the underlying AWS HPC technologies, including EC2, VPC, EFS, FSx for Lustre, ParallelCluster, Batch. Participants will have the opportunity to gain hands-on experience with various sample use cases.

ML – Seismology – Session Summary:

This workshop introduces participants to the ML/AI related services provided by AWS. It provides a gradual experience from domain-specific tools where no ML expertise is required to no-code ML (Amazon SageMaker Canvas) and learning tools (Amazon SageMaker Studio Lab). After a brief introduction of the Amazon SageMaker platform, participants familiar with Python development will be able to explore advanced self-paced examples of popular algorithms and ML frameworks, such as Tensorflow, PyTorch, MXnet.

Agenda:

9:00-9:30 am Check in and networking.
9:30-10 am Welcome and introductions
10am-12pm HPC Workshop - Time Domain Astronomy
12pm-1pm Lunch provided by AWS
1pm-3pm ML Workshop - Seismology
3pm-3:30pm Wrap-up and follow-up discussion

Note:

Faculty, research staff, postdocs and graduate students are welcome to attend from any discipline, although examples will be focused in seismology and astronomy.

For the astronomy (HPC) session attendees will benefit from familiarity with HPC clusters, ssh, command line, batch schedulers.

For the seismology (ML) session, attendees would benefit from familiarity with Python development and mainstream algorithms and ML frameworks, such as Tensorflow, PyTorch, MXnet.

Attendees MUST bring their own laptop/charger to participate in the hands-on portions and may (optionally) bring their own public (small) datasets to explore/experiment with. AWS will provide demo accounts at no cost to participants for the entire day.

All attendees must register here.