Above: generated sound – Below: original bird song.
A computer tirelessly makes attempts at pleasing a canary through sequences of synthesized sounds, a 3D-printer takes notes.
Initially six random sequences are generated, one for each lever in the canary’s cage. Based on the bird’s choice six new sequences are generated, the bird decides on the meme’s fitness for reproduction. The next day six algorithmically bred new sequences replace the old ones, only for the process to repeat itself. Hoping to find the canary’s ideal meme, through mimicking evolutionary principles.
Is a memetic algorithm the same as a genetic one? One big change I made in the genetic algorithm to fit for the memetic purpose is to generate a variable length gene or meme. Within the gene, a parameter decides on the gene’s length, within the mutating and breeding function those are dynamic. Making the gene~meme more like a function than a variable. Apparently, some other people were looking into it too. You can find the code for the project here, be warned tho, it’s quite crude. (Sonic-pi for sound, Repetier host to control the printer, works on raspberry pi with buttons)
This project is an effort towards an answer and an exploration in the nonhuman meme sphere:
- a study on bird song and nonhuman narrativity,
- an analogy to ubiquitous echo chambers,
- a manifestation of anthropomorphic naiveté,
- redefining training – be it an animal or algorithm,
- a cybernetic organ.
Please use this code, this idea for your own purposes, nest it, breed it, loop it.
A work by Floor Vanden Berghe