User:Alexander Roidl/thesis-outline4: Difference between revisions
(→Scope) |
(→Body) |
||
(4 intermediate revisions by the same user not shown) | |||
Line 7: | Line 7: | ||
====Argument==== | ====Argument==== | ||
Software Art can provide useful | Software Art can provide a useful framework and methods to approach machine learning algorithms artistically | ||
====Scope==== | ====Scope==== | ||
Line 21: | Line 21: | ||
==Body== | ==Body== | ||
===Drawing a parallel between Software Arts and the current state of machine learning=== | |||
====Software Art and Generative Art==== | |||
Argument: Software Arts can be seen as a reaction to the limited interaction with the tools in Generative Arts | |||
* History of Software Art & Generative Art | |||
* Software Art and its process based approach | |||
* Software as the main subject of research | |||
====Software Art applied to machine learning==== | |||
Argument: Methods and ideas of Software Arts and Software Studies can be applied to Machine Learning Algorithms and Software to create a more meaningful and lasting research. | |||
* Making machine learning the main field of research and the algorithm itself part of the art | |||
* Artists are not yet interacting with the deeper source -> how can the algorithm itself be the art? | |||
* Process based machine learning | |||
** What would be an intervention in the spirit of Software Arts? | |||
* Sidenotes: Code Brutalism | |||
===Learning Algorithms=== | ===Learning Algorithms=== | ||
====Machine Learning in Engineering==== | ====Machine Learning in Engineering==== | ||
Argument: Programers engage with machine learning algorithms in a very pragmatic way. Why it can be useful to | Argument: Programers engage with machine learning algorithms in a very pragmatic way. Why it can be useful to free those algorithms from their usefulness. | ||
* Example Language: How engineers talk about machine learning. Talking about features, generalisation, over and underfitting, training-data | * Example Language: How engineers talk about machine learning. Talking about features, generalisation, over and underfitting, training-data | ||
* Example Ideas: Usefulness and Speed as primary goals of development. | * Example Ideas: Usefulness and Speed as primary goals of development. | ||
* Example Resources: Only the fastest GPUS are used, who can’t afford it or who doesn’t have the data is out of the game. | * Example Resources: Only the fastest GPUS are used, who can’t afford it or who doesn’t have the data is out of the game. | ||
====Machine Learning in Art==== | ====Machine Learning in Art==== | ||
Argument: Current artistic use of machine learning is rather unsatisfying. Mostly these algorithms are used as tools to produce surprising outcome, but there is no artistic engagement with the algorithm itself. | Argument: Current artistic use of machine learning is rather unsatisfying. Mostly these algorithms are used as tools to produce surprising outcome, but there is no artistic engagement with the algorithm itself. | ||
Line 36: | Line 53: | ||
* Learning as a concept of loops | * Learning as a concept of loops | ||
https://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist Could this be an (unwanted) example for performative machine learning art? | https://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist Could this be an (unwanted) example for performative machine learning art? |
Latest revision as of 12:24, 10 January 2019
Thesis Outline
Thesis Statement
Topic
Learning Algorithms
Focus
a Software Art approach
Argument
Software Art can provide a useful framework and methods to approach machine learning algorithms artistically
Scope
The process oriented approach of Software Arts provides a useful theory that can be applied to machine learning algorithms in order to engage with them as an artist. The ideas and the deep investigation on the algorithms help to question and analyse the concepts of machine learning.
This refers mostly to the current use of machine learning and how this hype is used by artists to generate more or less random art using prewritten algorithms. And also how it is referred to as the magical tool for everything, providing surprising outputs with less effort. But still I think that there is something that can be learned from those kind algorithms and how arts can approach it in a more meaningful and lasting way.
I want to write an analytical essay that will relate to my practical work and build its theoretical foundation.
Introduction
Background
Recently new kinds of algorithms found their way into digital systems: Machine Learning Algorithms. They are increasingly implemented in our daily life and perform better than other kinds of previous system. At the same time machine learning raises other questions regarding e.g. opacity, bias and moral. Following this huge current attention machine learning is also being used in the arts. While recent artistic approaches are limited in their engagement with the technological understanding, I want to investigate the use of machine learning in a more sustainable way, asking what can be left after the hype of machine learning. How can we understand those algorithms as a object of studies or a piece of art itself?
Body
Drawing a parallel between Software Arts and the current state of machine learning
Software Art and Generative Art
Argument: Software Arts can be seen as a reaction to the limited interaction with the tools in Generative Arts
- History of Software Art & Generative Art
- Software Art and its process based approach
- Software as the main subject of research
Software Art applied to machine learning
Argument: Methods and ideas of Software Arts and Software Studies can be applied to Machine Learning Algorithms and Software to create a more meaningful and lasting research.
- Making machine learning the main field of research and the algorithm itself part of the art
- Artists are not yet interacting with the deeper source -> how can the algorithm itself be the art?
- Process based machine learning
- What would be an intervention in the spirit of Software Arts?
- Sidenotes: Code Brutalism
Learning Algorithms
Machine Learning in Engineering
Argument: Programers engage with machine learning algorithms in a very pragmatic way. Why it can be useful to free those algorithms from their usefulness.
- Example Language: How engineers talk about machine learning. Talking about features, generalisation, over and underfitting, training-data
- Example Ideas: Usefulness and Speed as primary goals of development.
- Example Resources: Only the fastest GPUS are used, who can’t afford it or who doesn’t have the data is out of the game.
Machine Learning in Art
Argument: Current artistic use of machine learning is rather unsatisfying. Mostly these algorithms are used as tools to produce surprising outcome, but there is no artistic engagement with the algorithm itself.
- Examples of current machine learning art
Neural Networks in Depth
Argument: A close up look at the algorithm reveals important key concepts
- The architecture / structure of learning algorithms
- Learning as a concept of loops
https://www.theverge.com/2016/3/24/11297050/tay-microsoft-chatbot-racist Could this be an (unwanted) example for performative machine learning art?
Conclusion
How the investigation and intervention in the process of neural networks can make artistic expression possible and comment on current tendencies in the field.