A lizard who has been given regenerative powers from the new adversarial cells
This is the code and experiments of the paper Adversarial Takeover of Neural Cellular Automata, published in ALIFE 2022.
We added adversaries to a Neural Cellular Automaton in order to change its color, shape or even both!
We have created a page with the video explanation and all the resources to learn about the work and experiments.
To use this package you just have to clone the repository:
git clone https://github.com/LetteraUnica/neural_cellular_automata
After cloning, execute
python3 -m pip install -r requirements.txt
which will install all the required packages to run the code.
Extra videos and resources can be found here
This work was inspired and/or helped by the following works:
- Alexander Mordvintsev et al. “Growing Neural Cellular Automata”. In: Distill (2020). https://distill.pub/2020/growing-ca.doi: 10.23915/distill.000232.
- Ettore Randazzo et al. “Adversarial Reprogramming of Neural Cellular Automata”. In: Distill (2021). https://distill.pub/selforg/2021/adversarial.doi: 10.23915/distill.00027.0043.
- Francesco Berto and Jacopo Tagliabue. “Cellular Automata”. In: The Stanford Encyclopedia of Philosophy. Ed. by Edward N. Zalta. Spring 2021. Metaphysics Research Lab, Stanford University, 2021.5.
- William Gilpin. “Cellular automata as convolutional neural networks”. In: Physical Review E 100.3 (Sept. 2019). issn: 2470-0053. doi: 10.1103/ physreve.100.032402. url: http://dx.doi.org/10.1103/PhysRevE . 100.032402.6.
- What Bodies Think About: Bioelectric Computation Outside the Nervous System - NeurIPS 2018 https://www.youtube.com/watch?v=RjD1aLm4Thg