29 October 2020
By: Jake Nylund and Daniel Slutsky

November 2020 machine learning study meetings

TL;DR: please register

The Clojure data science ecosystem changing rapidly, and it looks really promising at the moment. To support the coming developments, it is important to learn the current state of the ecosystem, to try it out in different contexts, to think about it, and to discuss its future directions.

Study meetings are one of the places where we will be doing all that.

Around the middle of November, we will meet to study some parts of the emerging stack for "classical" machine learning (as opposed to deep learning).

Below are a few details about the planned sessions.

Please register in advance as much as possible, and let us know if anything changes about your plans. This will help a lot in the preparations.

See you there!

Jake Nylund, Daniel Slutsky

Q: Can I go to any of the meetings?

A: Yes. Each of the mid-November meetings is self contained. They will all be similar in nature, and may overlap in content. You may join one or more of them, whichever you prefer.

Q: What should I do in order to participate?

A:

Q: What will happen in the meetings?

A:

  • We will begin with a short intro to some of the relevant libraries.
  • Then, participants will be invited to briefly share their screen if they have anything to show, ask or demonstrate.
  • Then, in the main part of the meeting, we will look into a data problem together and start tackling it. At any stage, someone will be sharing their screen with their REPL session, and the group will think together about the session.

Q: When will the meetings take place?

A:

Q: How long are the meetings?

A:

  • Each of the the mid-November study meetings will be 2 hours long.

Q: Are the meetings beginner-friendly?

A:

  • Not exactly. At the moment, we are studying tools and libraries which are still changing and breaking. Sometimes, they are not completely documented. Looking into them may require an open mind and might not be easy.
  • However, we do assume that nobody in the meetings is an expert. We will seek clarity and will make an effort to help each other.
  • After the ecosystem becomes more complete, we will be able to organize workshops which are much more beginner-friendly.

Q: What knowledge will be assumed?

A:

  • We will assume basic knowledge of Clojure (say, at least chapters 1,3,4,5 of Daniel Higginbotham's "Clojure for the Brave and True").
  • We will assume some basic understanding of machine learning concepts, mainly around typical settings of supervised learning. If you are not sure about it, we recommend to read the Wikipedia entry carefully, and discuss with us anything that you wish to understand better. If you have any doubt, let us talk.

Q: What platform will we use for the meetings?

A:

  • We will use Zoom for the video meetings. You will get a calendar invitation with a Zoom link after registration.
  • We will use the Clojurians Zulip chat for our notes and textual discussions. Here is some recommended background about our use of Zulip: https://scicloj.github.io/pages/chat_streams/

Q: What libraries will we study?

A:

  • According to our tentative plan, we will use tech.ml, fastmath and Smile for machine learning, tablecloth for data processing, tech.viz for data visualization, and Notespace or Oz for literate programming.

Q: How can I prepare for the meeting?

A:

  • We recommend trying out the relevant libraries, playing with them in the context of a data problem, and discussing what you find at Zulip.
  • We will share some recommended materials towards the meetings. Please follow the Zulip conversations.
Tags: meeting