Multidimensional Cellular Automata using glsl for modeling


Codex Virtualis is an artistic research framework oriented towards the generation of an evolving taxonomic collection of hybrid bacterial-AI organisms. With a subtle echo to the endosymbiotic (a) theory, we propose a symbolic formulation of a style transfer machine learning environment as a host, in which to merge bacterial/archaea time-lapse microscopy footage along with multidimensional cellular automata computational models (b) as endosymbionts, all under the orchestration of an autonomous generative non-adversarial network architecture (c). We aim as a result, to encounter novel algorithmically-driven aesthetic representations, tagged with a unique morphotype and genotype like encoding, and articulated around a speculative narrative encompassing unconventional origins of life on earth and elsewhere.

The project stands on the idea of cooperation to expand on the concept of intelligence by including machine and non-human agents into its configuration. And intends to articulate new schemes for the social imaginary to picture life outside the planet, and to better appreciate life on earth.

How can complexity outburst from simple dynamical systems when represented using convolutional neural networks?

What forms can arise from training creative algorithms with pattern-forming bacteria in spatially explicit environments and extremophile morphology modeling?

Two possible outcomes for presenting this project, which will be determined by the Covid-19 situation are: a. A site-specific immersive visual installation. b. An interactive online web/mobile site/app. In both cases displaying the complete organism collection, extensive process description, and DIY tutorials for people to replicate in a citizen science manner.

a. b. c.

Our motivation when applying for this residency resides in the possibility to incorporate our in-progress art & science project Speculative Communications [a], with research on extremophiles and complexity modeling studied at Seti. We are planning to create an image database by combining our collected visual data from bacterial colony growth with extremophile visual data and mathematical models of unicellular life forms collected at Seti, to pre-train a generative non-adversarial network, through style transfer interpolated from the biotic database and the self-organizing autonomous patterns into a continuous cellular automata. We are open for suggestions on other input sources to explore.
1. Map to represent the training model and the feedback loop created with the computer vision system.
We seek to activate a self-generative system, Artificial Intelligence, and algorithmic approximation for generating virtual organism models based on morphology and rule modeling. The system includes a sensorial element, via Computer Vision, which increases its ability to influence the training results by extracting data (morphology, mobility, etc) from each emerging generation. A feedback loop that might add on or remove rules from the cellular automata depending on the behavior of the Generative non -adversarial network. Stacking parameters into a continuous flow of integrative rules to stimulate evolution into the organism models Some of the essential research questions that we would like to tackle with this project are: how can we generate automated systems focused on the creation of new-to-nature organism models that inhabit the threshold between the biotic and the virtual domain?
2. Computer vision algorithms allow us to extract and encode visual, morphological and behavioral information. By following the correspondence between phenotype expression and the genotype-like encoding, we expect to direct some specific features of virtual organisms.
The novelty of this development resides in the cross-pollination of algorithms to stimulate a self-generative process and the possibility of influencing a generative non-adversarial network through the interaction between our computer vision (sensorial system) and the behavior of the cellular automata.
3. (a) bacterial behavior mosaic. (b) continuous cellular automata. (a) algorithmic thresholds for latent space.
We believe this system facilitates not only exploring emergencies between biotic and virtual but, over time, it can also be conditioned to create more specific experiments where we can address questions related to coevolutionary processes. As well as possible explorations on the environmental conditioning on the cellular automata to analyze interactions coming from extremophile emergencies. The three primordial elements of the full system are described in the following flowchart.

  1. The sources: from which we take characteristics biotic and virtual.
  2. The feedback: to retro control the cellular automata by integration of new rules.
  3. The virtual life generated: represented in a taxonomic way.
4 System model v.01.