GMU:Artists Lab:Theresa Elstner: Difference between revisions

From Medien Wiki
 
(7 intermediate revisions by the same user not shown)
Line 1: Line 1:
==Artists Lab 17/18==
==Artists Lab 17/18==
===Initial Post===
===Initial Post===
In this project I want to take a closer look on machine learning. I am interested in this method
Machine Learning is ubiquitous. Machines and gadgets all around us are continuously learning from the behaviour of their users and the surrounding world. I am interested in this learning process, its algorithmic mechanics and its motivation. I plan to concentrate my work on machine learning
*as an interactive tool,
*as an interactive tool,
*in terms of borders of possible applications (empathy, self-awareness)
*in terms of borders of possible applications (empathy, self-awareness),
*and in the fact that it can detect patterns, where we do not see any.  
*in the fact that it can detect patterns, where we do not see any
*and its use of "statistical stereotypes" for decision-making.


The wish for an intelligent system is very intriguing to me, because it raises the question, if behind this wish stands a socially disappointed human who wants to improve human interaction by replacing it on one side with a machine, ''who'' could maybe someday be a better companion than another human.  
[[File:Pixelatedportrait.png|link=MediaWiki|thumb|600px|Step back and you see a person in the picture. An ai could see this without stepping back.|center]]


[[File:Pixelatedportrait.png|500px|link=MediaWiki|thumb|Step back and you see a person in the picture. An ai could see this without stepping back.|center]]
[[File:Polygonshredder.gif|frame|link=MediaWiki|Jaume Sanchez' Polygonshredder seems to be alive when continuously forming new structures in chaos.|left]]
 
<br clear=all>
===Progress===
I am currently in a research phase, this includes
*theory: working on [https://www.uni-weimar.de/en/media/chairs/computer-science-and-media/webis/teaching/lecturenotes/#machine-learning lecture] and [https://www.uni-weimar.de/en/media/chairs/computer-science-and-media/webis/teaching/ws-201718/machine-learning/ exercises] by Benno Stein on Machine Learning
*application: playing around with [http://www.wekinator.orgwekinator wekinator] and this [https://github.com/hughrawlinson/wekinator-node helpful framework] for interfacing it with osc-protocol
 
The following picture shows the setup on my computer, while playing around with wekinator: It shows the Wekinator-UI on the upper right corner, some code to interface it with osc-protocol on the lower right corner, the voice-input-interface from the [http://www.wekinator.org/examples/ wekinator example set] on the lower left corner and finally the command-line-output of the classified voice on the upper left corner. The command-line-interface shows the output of a model, that is trained to classify 2 different voices. It outputs '1' for one voice, and '2' for the other voice.
[[File:Voicediscrimination.png|link=MediaWiki|thumb|700px|The model can distinguish two different voices.|center]]

Latest revision as of 20:40, 4 December 2017

Artists Lab 17/18

Initial Post

Machine Learning is ubiquitous. Machines and gadgets all around us are continuously learning from the behaviour of their users and the surrounding world. I am interested in this learning process, its algorithmic mechanics and its motivation. I plan to concentrate my work on machine learning

  • as an interactive tool,
  • in terms of borders of possible applications (empathy, self-awareness),
  • in the fact that it can detect patterns, where we do not see any
  • and its use of "statistical stereotypes" for decision-making.
Step back and you see a person in the picture. An ai could see this without stepping back.
Jaume Sanchez' Polygonshredder seems to be alive when continuously forming new structures in chaos.


Progress

I am currently in a research phase, this includes

The following picture shows the setup on my computer, while playing around with wekinator: It shows the Wekinator-UI on the upper right corner, some code to interface it with osc-protocol on the lower right corner, the voice-input-interface from the wekinator example set on the lower left corner and finally the command-line-output of the classified voice on the upper left corner. The command-line-interface shows the output of a model, that is trained to classify 2 different voices. It outputs '1' for one voice, and '2' for the other voice.

The model can distinguish two different voices.