# DeepLearning540 Technical Assistants Briefing

present:

- Peter Steinbach
- Erik Buhmann
- Helene
- Mykyta Shchedrolosiev
- Moritz
- Sascha Diefenbacher
- Manuel Sommerhalder
- Lucas Wiens
- Jonas Rübenach
- Tobias Lösche
- Oleg Filatov

## Icebreaker

### What was the last ML tool that you used?

- PS: a neural spline flow integrated in [`sbi`](github.com/mackelab/sbi/)
- Helene: anomaly_detection tool I wrote myself in python
- Moritz: pytorch
- Lucas: Keras, Tensorflow (long ago)
- Mykyta: Tensorflow and c++
- Erik: pytorch (for GANs)
- Sascha: PyTorch (for generative shower simulation)
- Manuel: tensorflow_probability
- Jonas: Tensorflow
- Tobias: PyTorch (normalizing flows)
- Oleg: PyTorch (transformers)
- William: tensorflow

### What was the ML algorithm that you learned ML with?

- PS: a CNN in keras to classify images, I think it was on the MNIST dataset IIRC
- Lucas: multiple small things in a lecture (e.g. image recognition)
- Helene: Linear Regression in High school
- Moritz: Linear Regression
- Jonas: Simple dense network to classify events
- Erik: Energy regression task with Keras
- Sascha: Keras for Top Tagging
- Manuel: Keras for classification
- Tobias: Simple fully connected network in TensorFlow/Keras
- Oleg: linear regression on some dummy dataset
- Mykyta: CNN for particle clasification
-

## Our Learners

### Data Science

You are provided with a CSV file. The file has 35000 rows. The file has 45 columns. Each column is separated by the "," symbol from the other. You would like to open the file in python, calculate the arithmetic mean, the minimum and maximum of column number 5, 12 and 39.

1. I can do that. Give me something that understands python and I'll show you.
+1
+1
+1
+1
+1

2. I'd need to look up the syntax in a cheatsheet or some old code and I'm good to do this.
+1
+1
+1
+1
+1
3. I am unclear about this, I'd have to consult a colleague or a search engine to do this.

4. I am not sure what to do.

### ML

You are helping to organize a conference of more than 1000 attendants. All participants have already paid and are expecting to pick up their conference t-shirt on the first day. Your team is in shock as it discovers that t-shirt sizes have not been recorded during online registration. However, all participants were asked to provide their age, gender, body height and weight. To help out, you sit down to write a python script that predicts the t-shirt size for each participant using a clustering algorithm.

1. I can do that. Give me something that understands python and I'll show you.

2. I'd need to look up the syntax in a cheatsheet or some old code and I'm good to do this.

3. I am unclear about this, I'd have to consult a colleague or a search engine to do this.
+1
+1
+1
+1
+1
+1
+1 never done clustering
+1

4. I am not sure what to do.
-why do clustering?

## Teaching / Infrastructure

- rule of thumb for teaching: 20/60/20
- the content that you teach is too slow for 20% of the learners
- the content that you teach is right for 60% of the learners
- the content that you teach is too fast for 20% of the learners

- inverted classroom setup
- goal: teach data science, introduction to ML
- 9x60 minutes time available in total
- 100 people/learners
- for each of the 9 lessons, I provide the following:
- a video (prerecorded) that learners can watch, 30-35 minutes
- lectures are based on colab notebooks
- check-yourself questions on the content of the video lecture
- practicals for about 20 minutes of time, derived/starting from the notebooks of the lecture
- essential: learners put taught content to use immediately
- split up the learners into 10 groups, assign each group to a assistant
- I will need all assistants from Monday, Mar 1, 2021 9-12am; the same on Tue+Wed
- if you already know that you there is an important appointment that you have to take
- either look for someone to help you, to replace you
- or let us know NOW that you cannot help
- mark yourself below here, when you cannot help:
- Erik, Tue 10-12am
- William, Monday 01.03 9-17

- infrastructure
- central information hub: https://deeplearning540.github.io
- content will be filled on this website
- videos will be uploaded to the indico agenda: https://indico.desy.de/event/28296/timetable/#20210301
- one zoom room per learner group and TA
- one hackmd document per learner group
- one central/global hackmd document for all
- chat tool?
- github account?
- +1
- +1
- +1
- +1
- +1
- +1
- +1
- +1
- +1
- Helmholtz/desy account?
- +1
- +1
- +1
- +1
- +1
- +1
- +1
- +1
- +1
- slack account?
- +1
- +1
- +1
- +1
- +1
- https://mattermost.web.cern.ch/
- Moritz + William will provide a channel on CERN mattermost for all assistants/instructors and organizers
1. Signup Cern Team:
- https://mattermost.web.cern.ch/signup_user_complete/?id=6sgzqof8u7yr8cik798ss5z34c
2. Instructors Channel
- Just write me (mscham0) a message once you are in the team.

- Erik: where will content appear
- watch this repo https://github.com/deeplearning540/deeplearning540.github.io
- PS will notify learners in mattermost channel about updates too (likely at the end of today)