Abstract:In evolutionary models, mutations are exogenously introduced by the modeler, rather than endogenously introduced by the replicator itself. We present a new deep-learning based computational model, the self-replicating artificial neural network (SeRANN). We train it to (i) copy its own genotype, like a biological organism, which introduces endogenous spontaneous mutations; and (ii) simultaneously perform a classification task that determines its fertility. Evolving 1,000 SeRANNs for 6,000 generations, we observed various evolutionary phenomena such as adaptation, clonal interference, epistasis, and evolution of both the mutation rate and the distribution of fitness effects of new mutations. Our results demonstrate that universal evolutionary phenomena can naturally emerge in a self-replicator model when both selection and mutation are implicit and endogenous. We therefore suggest that SeRANN can be applied to explore and test various evolutionary dynamics and hypotheses. Computational self-replicators are often modeled with implicit definitions of fitness, using various approaches such as biophysical modelling, molecular dynamics, ecological dynamics, and computational logic. However, mutation is usually explicitly defined using standard probability distributions such as the Poisson distribution. Here, we attempted to develop a computational self-replicator with implicit fitness and mutation processes using artificial neural networks, which are popular, complex, and flexible models for approximating solutions to a wide range of computational problems. The result is the self-replicating artificial neural network (SeRANN), which learns to copy its own genotype while simultaneously solving a computational problem such as image classification. Approximation errors implicitly introduce mutations, which affect the source code of the networks, changing network hyper-parameters (e.g, number of neurons in a layer), network architecture (connections between layers), and hyper-parameters of the learning algorithm. We evolved a population of SeRANNs and observed various evolutionary phenomena often seen in evolutionary experiments with microbes, demonstrating that this new evolutionary framework provides a promising model for further studies in evolutionary theory.

Self-Replicating Neural Programs

Self-replicating artificial neural networks give rise to universal evolutionary dynamics

Black-Box Ripper: Copying black-box models using generative evolutionary algorithms

Self-Replication, Spontaneous Mutations, and Exponential Genetic Drift in Neural Cellular Automata

Towards Self-Assembling Artificial Neural Networks through Neural Developmental Programs

Neuroevolution of Recurrent Architectures on Control Tasks

Hierarchical Self-Programming in Recurrent Neural Networks

Generative Adversarial Neuroevolution for Control Behaviour Imitation

Self-Programming Artificial Intelligence Using Code-Generating Language Models

A Self-Replication Basis for Designing Complex Agents

Correlation Analysis-Based Neural Network Self-Organizing Genetic Evolutionary Algorithm

Computational Life: How Well-formed, Self-replicating Programs Emerge from Simple Interaction

Recurrent Neural Network ODE Output for Classification Problems Follows the Replicator Dynamics

Born to Learn: the Inspiration, Progress, and Future of Evolved Plastic Artificial Neural Networks

Mechanical Self-replication

Neurocoder: Learning General-Purpose Computation Using Stored Neural Programs

Neural network design based on evolutionary programming

Exploration Of A Mechanism To Form Bionic, Self-Growing And Self-Organizing Neural Network

Replicator Equation: Applications Revisited

The Mating Rituals of Deep Neural Networks: Learning Compact Feature Representations through Sexual Evolutionary Synthesis

A theory of neural emulators