Exercise Session of 2 December, 2013

Submit your answers (at most one page per task, in separate pdf files) by Mon, Dec 2, 10.10 am in Moodle. Submit a pdf file for each task, of at most one page, discussing your solution/results/observations etc. If the task involves coding, also submit a file with the source code.

Scoring: 1 point for each completed task, at most 3 points.

Task 1 Music synthesis with SuperCollider

Install SuperCollider: http://supercollider.sourceforge.net/downloads/

Familiarize yourself with short code examples found on the page

http://ia600406.us.archive.org/16/items/sc140/sc140_sourcecode.txt. Try to modify and combine ideas presented in these examples to produce (a) a happy piece of music and (b) a sad piece of music. Which characteristics of music are associated with happy / sad emotions? The pieces should be (very) small in terms of the number of code characters.

Task 2 Probabilistic melody generation in SuperCollider

Try out the code snippet below. What does it do? Can you make this example to produce more interesting melodies by extending the set of different notes and defining the probabilities in a new way. You can also try including conditional probabilities for the notes following a certain note, i.e., specifying a Markov chain. You can find several SuperCollider tutorials on the internet by googling. This one provides information about patterns:

http://distractionandnonsense.com/sc/A_Practical_Guide_to_Patterns.pdf

( Pbind(

           \degree, Pwrand((1..3), [1,2,8].normalizeSum, inf),

       \dur, 0.5

).play;

)

Task 3  Fitness Function Generation for Visual Creativity

Read the paper

Simon Colton. Automatic invention of fitness functions with application to scene generation. Applications of Evolutionary Computing. Springer Berlin Heidelberg, 2008. 381-391.

Consider Cityscape Inspired by the Style of Salvador Dalí. In this task the idea is to create the “flow-like” feeling as seen in the image above. Define the scene, scene elements, scene element properties, possible correlations between scene elements, possible correlations between the properties of a scene element. Define a fitness function which is going to be used for scene generation.

Task 4  Social creativity and metacreativity

Consider the agent you implement(ed) for the project (or think of some other creative agent, specify which). Outline the agent briefly by using the model of creativity as search (Wiggins).

1. Does your agent potentially suffer from the following? Why/why not? Also give concrete examples of concepts (artefacts) using your implementation.

1. generative uninspiration
2. pointless aberration
3. productive aberration

2. Is your agent able to benefit from social interaction with other agents to solve the above problems? How? Or what would it take to extent your agent to do so?

3. Can you think of other potential problems with respect to the cooperation of R, E, and T?