AI, ML, & Distributed Computing at rstudio::global(2021)

Here's a recap of some of the activity related to AI, ML, and Distributed Computing at rstudio::global(2021).

BTW if you aren't signed up yet ... please follow RStudio's AI Blog to get all the latest news from the multiverse (a.k.a mlverse) team!

And if you haven't yet and feel like, it isn't too late yet to take part in our team's feedback survey - see below :slight_smile:


Welcome

Sigrid Keydana to AI / Machine Learning - January 21st, 5:48 AM

Hi there, welcome everyone to the channel dedicated to all topics big data & AI: Spark & sparklyr, deep learning with torch and/or tensorflow/keras, and more!

Here, among many active community members, contributors, and enthusiasts, you'll meet the RStudio multiverse team: Yitao, Daniel, and me (Sigrid). Let's have fun and interesting conversations!

I also have something to ask you :slight_smile:

We've created a survey that should help us find out more about how we can serve our community. The survey is about where/how you use (or would like to use) sparklyr, torch, and tensorflow, what we could do better, and any ideas and wishes you have.

It is completely anonymous, we're not interested in tracking any personal information. Please do us a favor and take part! Here is the link:

https://surveyhero.com/c/multiverseb9483c64

Many thanks and see you soon!


ethics & surveillance in AI

Sigrid Keydana to AI / Machine Learning -[January 21st, 10:05 AM

Hi, as we were talking about ethics & surveillance in AI ... here are the Medium posts I mentioned :slight_smile:

Privacy / surveillance capitalism:

https://medium.com/@zkajdan/the-hard-problem-of-privacy-4bc0f731c16e

AI ethics is not an optimization problem:

https://medium.com/@zkajdan/ai-ethics-is-not-an-optimization-problem-fd927d8cb1a5

I am very happy to hear that others, too, care about these things and sincerely hope we can somehow, as a community, raise awareness.

Grant Fleming - January 21st, 11:43 AM
People also tend to get frustrated when issues around ethics/fairness/etc. in data science are discussed without any sort of actionable, technical advice being offered. Recent work is helping to make this less hand-wavey, for example, the framework discussed in Raji et al's "Closing the AI Accountability Gap" paper ([2001.00973] Closing the AI Accountability Gap: Defining an End-to-End Framework for Internal Algorithmic Auditing).
Also, unsubtle plug: I cover this exact topic in my talk later today, "Fairness and Data Science: Failures, Factors, and Futures", if anyone wants to discuss these topics in an R-specific context :smiley:


cloud platforms for using keras or torch

Chad Peltier to AI / Machine Learning - January 21st, 10:13 AM
Does anyone have opinions on cloud platforms for using keras or torch in R? i.e. in Paperspace or Sagemaker?

Daniel Falbel - January 21st, 10:24 AM
I had a reasonable experience with https://colab.to/r , paperspace gradient notebooks and https://nextjournal.com/ although I still prefer using RStudio server installed in an EC2 or GCE instance.

Adam Black - January 21st, 12:12 PM
Wow I had no idea we could use Google collab with R. That's cool.


new torch website

Sigrid Keydana to AI / Machine Learning - January 21st, 5:39 PM

In case you missed it on twitter, the torch website has a brand new "getting started" section - trying to address experienced Keras/TensorFlow users as well as users new to deep learning

3 Likes

This topic was automatically closed 42 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.