You're invited to join me for a Cross Roads talk next week, hosted by Cross Labs in Kyoto and streaming at https://t.co/Q3agn3NvH1.
I'll be talking about discretization and self-organization in cellular automata and similar systems.
In @crosslabstokyo talk, Tyrell showed a bunch of examples of complex systems that depend (feed) on discretization, including Lenia, SmoothLife, Neural CA, reaction-diffusion systems, and Lorentz attractor!
For anyone who missed the stream or wants to revisit the talk, the recording will remain up on YouTube. The slides contain additional examples we didn't get to in the talk, and are available at https://t.co/2ictYz6evD
👾 What would you do if you were suddenly transported to a discretized world? 👾
Next week, Q. Tyrell Davis will show us how a certain coarse fabric of reality is necessary for agents to emerge and persist.
Don't miss out! Aug 22 @ 10:30am JST 👉 https://t.co/BC64Letaom
You're invited to join me for a Cross Roads talk next week, hosted by Cross Labs in Kyoto and streaming at https://t.co/Q3agn3NvH1.
I'll be talking about discretization and self-organization in cellular automata and similar systems.
@ALifeConf Is there a list of all entrants? I have a playlist of the ones I could find (https://t.co/QAAwz3KGLp), but it was missing several of the winners.
Here is a playlist of the entries I could find for the ALife Virtual Creatures Competition:
https://t.co/tMD1FrLhpL
Feel free to reply with a link if I missed yours.
The update rule for this cellular automaton is based on adaptive moment estimation (adam) optimization. The RGB channels are grid states, second moment, first moment, respectively.
It shares a neighborhood kernel and morphological similarities with Orbium patterns from Lenia.
Some of my work on the relationship between the "necessary evil" of discretization and glider autopoiesis in nominally continuous cellular automata and similar systems, published in Innovations in Machine Intelligence, the journal of @crosslabstokyo https://t.co/k379eKahQ5
I came back to reaction diffusion systems and managed to get a Gray-Scott framework properly implemented.
From left to right: u-glider and Daedalus from https://t.co/9AgDDLO4Xi, and a potentially new pattern in U-Skate rules I'm calling "berry cup" #alife#cellularautomata
T-shirt designs developed using evolving #cellularautomata models of bio-pattern morphogenesis. https://t.co/h4fSuFk6Fo
https://t.co/nycXyMjPSB #generativeart
I'm considering spatial/temporal resolution and numerical precision. Each has one or more counterparts in NNs.
1. Numerical precision (quantization)
2. Spatial (kernel size)
3. Temporal. (Less obvious, but residual connections can generally be written as a step of Euler's method)
My current project examines the effects of discretization on self-organising patterns in #CellularAutomata and other glider-supporting systems. The effects can be counter-intuitive, e.g. instability due to _higher_ spatial resolution, as in the _Scutium_ glider from #Lenia below.
It reminds me of the "Face on Mars" from the Cydonia region of the red planet. When photographed at relatively low resolution by Viking 1, it's clearly a monumental artifact made by extraterrestrials. When imaged with a bit more resolution by HiRISE, it's obviously just a hill.