Adaptive Epigenetic Neural Network

The Story: In June 2025, I was trying to develop a machine learning project to fill out my portfolio. I ended up getting way too involved into the project, and developed an entire simulation around the concept that I am calling the Adaptive Epigenetic Neural Network. I developed these networks to work in a genetic life cycle algorithm where they would train, go through selection, and reproduce all with appropriate operators. The key difference between the neural networks in this projects and those in other similar algorithms is the addition of a genetic operator that learns through backpropagation along with the bias.

The Result: I was able to use my old laptop and github actions to automate the running and model evaluation for this simulation. My preliminary results gave me 97.90% classification accuracy on the Wisconsin Breast Cancer Dataset. Additionally, the success of this project inspired me to start livestreaming again and go full stream towards become and independent scientist. I made visuals for the simulation to try and make it more interesting to run the simulation in the background while I do other things on stream.

You can find my paper on the project here: https://osf.io/preprints/osf/mfu9w_v1

I was also featured on a podcast where I was able to talk about the project. The link to that can be found here: https://www.youtube.com/live/24NMRVbLJd8?si=rEDJwZtJj30I8BhT

Finally, I have also started a GoFundMe for Boston’s Children’s Hospital Blood and Cancer Disorder Center. I am doing this to try and produce a meaningful impact from this research project. The link to that can be found here: https://gofund.me/d8a5a20e