My project is to create neural networks that can evolve like living organisms. This mechanism of evolution is inspired by real-world biology and is heavily focused on biochemistry, which contribute to creating a complex gene regulatory network and an indirect encoding mechanism for neural networks, where even a single letter mutation can cause dramatic changes to a model.
GITHUB: GitHub - Danil-Kutnyy/Neuroevolution: Simulation of neural network evolution
The code for this project consists of three parts:
- Genpiler (a genetical compiler) - the heart of the evolution code, which simulates many known biochemistry processes of living organisms, transforming a sequence of “ACGT” letters (the genetic code) into a mature neural network with complex interconnections, defined matrix operations, activation functions, training parameters and meta parameters.
- Tensorflow_model.py transcribes the resulting neural network into a TensorFlow model.
- Population.py creates a population of neural networks, evaluates them with MNIST dataset and creates a new generation by taking the best-performed networks, recombining their genomes (through sexual reproduction) and mutating them.
This is very interesting work. Hats off to you! I was thinking about something similar and was formulating my own models for population evolution, but what you have done here is well beyond what I was imagining. The README file in your github repo is a useful explanation of your project. I hope you have great success with this!
Thank you very much! I really hope it will be useful. It’s just that I thought how cool this idea is, and it seems that nobody was doing anything close to such complex encodings. I’ve been reading neural network architecture evolution papers and everybody was doing strange things, why nobody copy biology! We know so much now, it’s crazy nobody tried this stuff! And I am crazy about it!
Wow, your project on creating neural networks that evolve like living organisms is truly fascinating! The use of biochemistry-inspired mechanisms and indirect encoding for neural networks showcases a unique approach. Your code implementation, including the Genpiler, TensorFlow model transcription, and population creation, demonstrates a comprehensive framework. I appreciate the thorough README file in your GitHub repository, providing valuable insights into your project. Your innovative idea and passion for incorporating biological principles into AI evolution are commendable. Best of luck with your endeavors!