Tutorials
NCWCP Conference Center, 3:00-5:00PM
Tutorial 1
- Title:
- How to do machine learning in the cloud: a fast intro to public data sets, Keras, TensorFlow, and Earth engine
- Instructor(s):
- Valliappa Lakshmanan
- Organization:
- Google
- Description:
- In this tutorial, you will get to work with the public cloud for scientific
data processing and machine learning: specifically how to access, publish, analyze,
and do machine learning on datasets at scale. You will learn how to work with Cloud Shell,
Notebook Instances and easily parallelize analysis and training tasks.
- Requirements:
- This tutorial will be hands-on. All participants will need their own computer.
- Tokens will be provided for this tutorial.
- Location:
- NCWCP Conference Center A/B
Tutorial 2
- Title:
- Deep Learning in Python for Environmental Prediction
- Instructor(s):
- David Gagne (NCAR) and Ryan Lagerquist (OU/CIMMS)
- Organization:
-
University of Oklahoma/Cooperative Institute for Mesoscale Meteorological Studies (CIMMS)
University of Oklahoma/National Center for Atmospheric Research (NCAR)
- Description:
- Participants will learn how to train a convolutional neural network (CNN),
a common type of deep-learning model, to make weather phredictions from gridded data.
They will also learn several interpretation methods for machine learning - including
permutation, backwards optimization, saliency maps, and class-activation maps - which
attempt to explain the model and understand the physical relationships it has learned.
- Requirements:
- This tutorial will be accessed through Jupyter notebooks on the cloud. All participants will need their own computer.
- Location:
- NCWCP Auditorium
Tutorial 3
- Title:
- Unleashing machine learning's potential impact through containerized deployment
- Instructor(s):
- Patrick Flickinger (CELA Data Science, AI for Earth Engineering)
- Organization:
- Microsoft Corporation
- Description:
- After developing an algorithm or machine learning model,
researchers face the problem of deploying their model for others to consume,
integrating it with data sources, securing its access, and keeping it up to date.
The many possible sources of data – from on-premises file shares, to cloud-based feeds,
to IoT devices – further complicate the utility and scalability of a model in the field.
Consequently, sophisticated models representing years of research often remain confined
to a researcher's computer, never reaching their full potential as practical solutions.
In this workshop, we demonstrate how a model developed on Microsoft's Azure cloud computing
platform can be turned into a scalable API, highlighting its interoperability with other
applications and systems. We will introduce Docker for containerizing the API, allowing a
consistent execution environment for the model, regardless of the hosting infrastructure,
how to push the resulting Docker image to a container registry, and how to deploy it on
Azure for worldwide access. Finally, we will examine how AI for Earth organizations have
used the framework to deploy impactful APIs and how some have taken advantage of its extensibility
to build self-scaling pipelines.
- Requirements:
- This tutorial will be hands-on. All participants will need their own computer.
- Updated links for tutorial will be shared with registered participants before the tutorial.
- Location:
- NCWCP Conference Center C
-
About the Tutorials
The Tutorial sessions at the NOAA Artificial Intelligence Workshop will provide a hands-on experience
with Artificial Intelligence and its applications.
Tutorials will take place simultaneously from 3:00pm-5:00pm, Thursday, 25 April 2019.
Tutorial Sign-up
The deadline to register is 19 April 2019.
Interested attendees should select from the three tutorials in order of their preference. Please note, there
are only 25 seats for each tutorial. This is first come, first serve.
Tutorial registration will be confirmed by 23 April 2019
Note to Foreign Nationals: Foreign Nationals approved for the event will be able
to access the Tutorial sessions
Contact:
For more information, please contact:
Narges Shahroudi
|