Excited to be in #AGU20’s tutorial on #MachineLearning for #Geoscience today with over 1200 registered participants from across the world!
First time I'm hearing about the Zarr data format for breaking large datasets into a large number of small binary files that are good for cloud storage. zarr.readthedocs.io/en/stable/
Thanks so much @DJGagneDos for a great intro ML talk at #AGU20 and for sharing a link to your recorded talk/slides as presented to the NCAR ARTIFICIAL INTELLIGENCE FOR EARTH SYSTEM SCIENCE (AI4ESS) Summer School! www2.cisl.ucar.edu/events/summer-…
Some additional really great ML fundamentals resources by way of Ankur Mahesh at Berkeley:

J. Shewchuck's "Concise ML" lecture notes: people.eecs.berkeley.edu/~jrs/jrspapers…

S. Nasiriany, G. Thomas, et al.'s "A Comprehensive Guide to Machine Learning": snasiriany.me/cs189/
It was helpful to see the concrete example of using neural nets to emulate the output from the Community Land Model by @katiedagon during the #AGU20 ML tutorial:

Looking forward to seeing your forthcoming paper in …matology-meteorology-oceanography.net
Lots of references are being made to prior meetings this year with online talks. One relevant to ML is: slideslive.com/climateinforma…
So many good questions at the #AGU20 ML tutorial, and Karthik Kashinath and the other conveners have been really quick to post relevant articles and blogs that cover a myriad of topics. TowardsDataScience.com seems to come up frequently.

For example, towardsdatascience.com/clearly-explai…
No surprise that under/over-fitting data causes poor performance in machine learning as it does with many aspects of research and life! Another good resource from machinelearningmastery.com mentioned at the ML tutorial.

How to choose a metric for evaluating uncertainty in machine learning? Some ideas: cs.adelaide.edu.au/~javen/talk/ML…
That time when you pull up a relevant and well written article on feature selection and reducing the dimensionality of data and you stumble upon a weird (and awesome!) data science competition to not overfit data.

Thinking about #AirborneScience applications for object tracking reminds me of the (apocryphal?) stories of grad students of yore having to watch 8-10 hours of forward cam videos per flight to identify each and every time the airplane flew through a cloud. missinglink.ai/guides/compute…
Tapping out early to go make dinner for the kiddos, but really enjoyed this afternoon's #AGU20 #MachineLearning tutorial! Thanks to all the conveners and presenters for an outstanding set of talks, resources, and practical examples! agu.confex.com/agu/fm20/meeti…

• • •

Missing some Tweet in this thread? You can try to force a refresh

Keep Current with Rich Moore

Rich Moore Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!


Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @richardhmoore

10 Dec
Just getting around to watching the many excellent #AGU20 talks on the atmospheric effects of COVID-19-driven societal changes. While I would've liked to have attended the live sessions, I am loving watching these on demand at 1.5-2x speed. A few highlights (of many) in a thread:
Greg Frost presents a scientific smorgasbord of @NOAA's COVID AQ research in the United States over recent months. Measurements and fuel-based inventories show significant reductions in motor vehicle emissions that are confirmed by satellites. agu.confex.com/agu/fm20/meeti…
Daniel Sauer (@dns4861) reports exciting results from the @DLR_en #BlueSky Falcon campaign that suggest the ~80% reduction in commercial flights over Germany translate into reduced particle and reactive traces gases concentrations in the upper troposphere. agu.confex.com/agu/fm20/meeti…
Read 12 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!