Welcome to Etherpad!

This pad text is synchronized as you type, so that everyone viewing this page sees the same text. This allows you to collaborate seamlessly on documents!

Get involved with Etherpad at http://etherpad.org
http://pad.constantvzw.org/public_pad/TouchingCorrelationsWorkshop


Hi Everyone,

First of all thank you for registering for the touching correlations workshop. Here is an email to provide you with an overview of our plans.

We will meet on Saturday and Sunday between 10am and 4pm at the premises of Data and Society [0]. You need to buzz us when you are downstairs so that we can let you in. In case you have problems, please give me a call: +13478667663.

This is going to be a collaborative workshop more than a structured leason plan so please come with questions and ideas to share!
Some of you informed us that you are trying to organize child care, please do let us know if this is still proving to be an obstacle.

We will organize some food. This means we invite you to help us put it together and clean up afterwards.

In preparation of the course, we would like you to install WEKA.

Video on how to Install WEKA:
    https://www.youtube.com/watch?v=nHm8otvMVTs     

If you have time to prepare more, here are some fun videos which provide an overview of machine learning.
    https://www.youtube.com/watch?v=-rMMTv7XLYw
    https://www.youtube.com/watch?v=EFrgVDniDqU

If you still have time, you may want to read the first chapter of the Data Mining Book that accompanies WEKA [1].
   
For those who already have an experience in computer science and machine learning, we would love it if you could take a more active role when it comes to technical stuff. Could you please send us an email, so that we can coordinate.

We are totally looking forward to having you with us. Have a nice week, we will have the weekend!
Courtenay and Seda

   
[0] http://www.datasociety.net
[1] http://bit.ly/1kWsnaO



SCRIPT:
    
Day 1:
    
10.00 - 10.30 settling in/getting coffee/bagels

10.30  introductions

• us:
  • who we are?
  • why did we do this?
    • - coz
    • - to take knowledge for a walk
    • - knowledge not being something that gets passed on but that is situated -> participatory workshop
    • - how do we co-exist in a world where machine learn, perceive, co-think, and articulate our daily lives?
      • "The knowledge generated by profiles is not static universal knowledge, but it is knowledge that is always in the process of being made"
      • i would say the same about machine learning: the science of machine learning is in the process of being made, and this session is an invitation to co-make it.
    • - it is popular (in our circles) and we talk about it a lot, let's touch it
    • - i know courtenay!!!
    • - we don't know everything, and since we are going to develop a new machine learning, we are in it together!
    • 
      
    • 
      
    • - I'm interested in having more popular understanding of ML
    • - interested in getting other perspectives on the field
    • 
      
• plan for the weekend 
  • time plan
  • etherpad
  • who installed successfully: we need a volunteer to help those who couldn't install
  • food (we cook together and we clean together)
  • toilettes
  • wifi
  • 
    
  • 
    
• who they are?
  • why they are there? where did they experience machine learning?
  • questions they are interested in?
  • 
    

12:00 - 13.30 

• courtenay runs through the slide deck
  • preface: please ask anything, throughout the weekend
  • take time/open discussion
• 
  
• 
  

13.30 - 14.30    

• lunch: some people prepare food
• some people make sure all the software is installed
• 
  
• 
  

14.30 - 16.00

• look at the explore interface of weka
• a dataset of courtenay's choice, e.g., glass arff: look at the arff file, the header, the formatting, discuss what is and isn't in the file
• my first classifier
• 
  

HOMEWORK: 

•     Find a dataset
  • give links to datasets
  • 
    
    • 
      

- maybe elicit initial ideas of what people think ML is / what experience they have with it?
- do (some of?) my slides
- show some of the youtube videos you found, the basic intro ones
- dive into weka (most of day 1?) -- how much do you think I need to pre-plan what I want them to explore here?

Day 2:


10.00 - 10.30 welcoming/arriving

10.30 - 12.00 on correlations/statistics

12.00 - 13.00 more WEKA: do some clustering and (if time permits) regression

13.00 - 14.00 Lunch

14.00 - 15.00 play with your own dataset (they can share/work in groups)

15.00 - 16.00 videos and general discussion

Videos:    
sentiment analysis: https://www.youtube.com/watch?v=ytUHvMNnzZk
advertising application: https://www.youtube.com/watch?v=EQhwNcQhP4g

Deep dream:
http://googleresearch.blogspot.com/2015/06/inceptionism-going-deeper-into-neural.html
http://motherboard.vice.com/read/why-googles-neural-networks-look-like-theyre-on-acid



To do!
ask data and society for coffee
overhead projector
sandy and carlin 
shopping list
prepare links to datasets available online
http://archive.ics.uci.edu/ml/

ask participants:

• to formulate questions: send them to us beforehand?
• install weka: tips on installing stuff
• let us know what operating system they are using?
• give them the first chapter of the dm book to read (very accessible/no math)
• 
  

some questions:

•     are data miners moving more and more towards unsupervised learning?
  • could learning be more emancipatory: we don't start with preconceived ideas of gender, race, class? what does that mean?
  • do data miners not necessarily care about inputs that provide "good results?
    • maybe your choice of browser says more about your employability than the rest of your cv? (an example by mark andrejevic.
    • or is this just the wrong way to think about algorithms: think of google's deep dream where the algorithms recognize weights always with the associated muscular arms.
    • 
      
  • i guess the more generalized question is: how can we understand the relationships between inputs and outputs in the various approaches to machine learning?

WEKA!


Installing Weka:
    https://www.youtube.com/watch?v=nHm8otvMVTs

Weka Exercises:

• Install
• Use explore, take a look
• Try some data sets
  • nominal vs. numeric data fields
  • look into arff files, find other datasets online and see their comments, what is discussed what is not discussed!
• Language used to notate the data sets: return as a new row. limitation on representation, how do you combine datasets? 
• looking into reasonable values:
  • "click around to make sure that things look real!"

Ongoing weka course (non-supported)
============================================
https://weka.waikato.ac.nz/dataminingwithweka/preview

The book:
    http://bit.ly/1kWsnaO
Material for Touching Correlations:

Videos:
    
    Introduction to Weka (from the makers!!!)
    https://www.youtube.com/watch?v=Exe4Dc8FmiM
    
    Stanford Lecture Series (including materials)
    https://www.youtube.com/watch?v=UzxYlbK2c7E
    http://www.stanford.edu/class/cs229/
    
    
Books:
    A forthcoming novel on artificial intelligence:
        http://www.harpercollins.com/9780062391193/speak/web-sampler
    

Deep Learning:
    
    Google's Deep Dream Project:

• 
  
  • Blog Post by Google: Inceptionism: Going deeper into Neural Networks
  • http://googleresearch.blogspot.co.uk/2015/06/inceptionism-going-deeper-into-neural.html

• Deep Dream Code on Github:
• https://github.com/google/deepdream
• http://googleresearch.blogspot.com/2015/07/deepdream-code-example-for-visualizing.html

• Yes, androids do dream of electric sheep
• http://www.theguardian.com/technology/2015/jun/18/google-image-recognition-neural-network-androids-dream-electric-sheep

• Google and Spam:
• http://www.csmonitor.com/Technology/2015/0713/Google-fights-spam-with-artificial-intelligence
• 
  

    Algorithms of the Mind: What machine learning teaches us about ourselves?
    https://medium.com/deep-learning-101/algorithms-of-the-mind-10eb13f61fc4

    Computer Science Paper: Deep Neural Networks are Easily Fooled: High Confidence Predictions for Unrecognizable Images

• http://arxiv.org/abs/1412.1897

Writings on On Algorithms:
     

• Louder vowels won’t get you laid, and other tales of spurious correlation
• http://arstechnica.com/science/2015/06/louder-vowels-wont-get-you-laid-and-other-tales-of-spurious-correlation-2/  

    

• Politique des Algorithmes (in French and in English)
• http://www.scoop.it/t/politique-des-algorithmes

    
Decision Tree Exercise:
    http://users.sussex.ac.uk/~christ/crs/ml/lab-exercises.html
    also http://www.saedsayad.com/decision_tree.htm



    Also see Cqrrelations workshop, which inspired "Touching Correlations". Many of the references above are also from the cqrrellations mailing list. 
    http://www.cqrrelations.constantvzw.org/1x0/
    
    for those of you who would like to work with them, they organize a yearly summer school called re-learn:
    http://relearn.be/2015/