Growing Neural Cellular Automata

Mordvintsev, Alexander; Randazzo, Ettore; Niklasson, Eyvind; Levin, Michael

doi:10.23915/distill.00023

Cited by 112 publications

(127 citation statements)

References 21 publications

Supporting

Mentioning

121

Contrasting

Unclassified

Order By: Relevance

“…The most straightforward variants simply alter the ruleset, such as the cellular automatons "Flock", in which cells can only survive with 1 or 2 neighbours, and "Day and Night", which has four rules for survival, and four rules for being born. Other variants introduce more fundamental changes, such as converting Life into 3D [3], changing the size of the neighbourhood (e.g., Larger than Life [8]), changing the grid to be non-square [4], and replacing the rules of Life by a neural network that learned to "regrow" certain patterns [11].…”

Section: Game Of Life Extensions and Variantsmentioning

confidence: 99%

Problife: a Probabilistic Game of Life

Vandevelde¹,

Vennekens²

2022

Preprint

View full text Add to dashboard Cite

This paper presents a probabilistic extension of the wellknown cellular automaton, Game of Life. In Game of Life, cells are placed in a grid and then watched as they evolve throughout subsequent generations, as dictated by the rules of the game. In our extension, called ProbLife, these rules now have probabilities associated with them. Instead of cells being either dead or alive, they are denoted by their chance to live. After presenting the rules of ProbLife and its underlying characteristics, we show a concrete implementation in ProbLog, a probabilistic logic programming system. We use this to generate different images, as a form of rule-based generative art.⋆ This research received funding from the Flemish Government under the "Onderzoeksprogramma Artificiële Intelligentie (AI) Vlaanderen" programme.

show abstract

Section: Game Of Life Extensions and Variantsmentioning

confidence: 99%

Problife: a Probabilistic Game of Life

Vandevelde¹,

Vennekens²

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…In fact, they show that any CA can be represented with a certain kind of CNN, and with an elegant demonstration of Conway's Game of Life 14 in a CNN, illustrating that in certain settings, CNNs can exhibit interesting self-organizing behaviors. Recently, several works such as Mordintsev et al 38 that we will discuss later have exploited the self-organizing properties of CNN, and have developed neural network-based cellular automata for applications such as image regeneration.…”

Section: Collective Intelligence For Deep Learningmentioning

confidence: 99%

“…Implicit relationships and recurring patterns in natural medium can benefit from employing approaches from cellular automata in learning alternative representations of images. Like CeNNs, the Neural Cellular Automata (neural CA) model proposed by Mordvintsev et al 38 treated each individual pixel of an image as a single neural network cell. The networks are trained to predict its color based on the states of its immediate neighbors, thereby developing a model of morphogenesis for image generation.…”

Section: Image Processingmentioning

confidence: 99%

“…The resulting systems have proved to be promising solutions to some of the problems mentioned above. For example, a cellular automata-based neural network is resistant to noise and can even "self-heal" when damaged 38 . As another example, modular, decentralized self-organizing controllers not only demonstrated the possibility of a general controller but also maintained performance when deployed on multiple robots of different morphology in simulation 29 .…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Collective Intelligence for Deep Learning: A Survey of Recent Developments

Ha¹,

Tang²

2021

Preprint

View full text Add to dashboard Cite

In the past decade, we have witnessed the rise of deep learning to dominate the field of artificial intelligence. Advances in artificial neural networks alongside corresponding advances in hardware accelerators with large memory capacity, together with the availability of large datasets enabled researchers and practitioners alike to train and deploy sophisticated neural network models that achieve state-of-the-art performance on tasks across several fields spanning computer vision, natural language processing, and reinforcement learning. However, as these neural networks become bigger, more complex, and more widely used, fundamental problems with current deep learning models become more apparent. State-of-the-art deep learning models are known to suffer from issues that range from poor robustness, inability to adapt to novel task settings, to requiring rigid and inflexible configuration assumptions. Ideas from collective intelligence, in particular concepts from complex systems such as self-organization, emergent behavior, swarm optimization, and cellular systems tend to produce solutions that are robust, adaptable, and have less rigid assumptions about the environment configuration. It is therefore natural to see these ideas incorporated into newer deep learning methods. In this review, we will provide a historical context of neural network research's involvement with complex systems, and highlight several active areas in modern deep learning research that incorporate the principles of collective intelligence to advance its current capabilities. To facilitate a bi-directional flow of ideas, we also discuss work that utilize modern deep learning models to help advance complex systems research. We hope this review can serve as a bridge between complex systems and deep learning communities to facilitate the cross pollination of ideas and foster new collaborations across disciplines.

show abstract

“…Introduced by John von Neumann and Stanislaw Ulam in the 1940s (Von Neumann, 1951), CA have continued to fascinate, as their conceptual simplicity serves as a powerful microscope that allows us to explore the nature of computation and complexity, and the origins of emergence. Far from being an antiquated computational model, investigators are utilizing CA in novel and creative ways, such as the incorporation with Deep Learning (Mordvintsev et al, 2020;Nichele & Molund, 2017). Popularized and investigated by Stephen Wolfram in his book A New Kind of Science (Wolfram, 2002), CA remain premier reminders of a common theme in the study of the physical world: that simple systems and rules can give rise to remarkable complexity.…”

mentioning

confidence: 99%

CellPyLib: A Python Library for working with Cellular Automata

Antunes¹

2021

JOSS

View full text Add to dashboard Cite

Growing Neural Cellular Automata

Cited by 112 publications

References 21 publications

Problife: a Probabilistic Game of Life

Problife: a Probabilistic Game of Life

Collective Intelligence for Deep Learning: A Survey of Recent Developments

CellPyLib: A Python Library for working with Cellular Automata

Contact Info

Product

Resources

About