berkedenizbozyigit/novel-artificial-neurons

What if neurons didn't just sum their inputs? An MSc research project reimagining neural aggregation with learnable F-Mean and Gaussian functions, showing improved noise robustness on CIFAR-10.

28
/ 100
Experimental

This project offers new types of artificial neurons for deep learning models, helping improve how neural networks process information. Instead of simple sums, these neurons use advanced, learnable mathematical functions (like F-Mean or Gaussian functions) to combine their inputs. The goal is to create more robust and effective image recognition systems, especially when dealing with noisy or imperfect data. Deep learning researchers and practitioners focused on computer vision or building more resilient AI models would find this useful.

No commits in the last 6 months.

Use this if you are developing deep learning models, particularly for image recognition tasks, and want to explore novel neuron architectures that offer improved noise robustness compared to traditional methods.

Not ideal if you are looking for a plug-and-play solution for general machine learning problems without delving into fundamental neural network architecture research.

deep-learning-research computer-vision neural-network-architecture noise-robustness artificial-intelligence
Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 6 / 25
Maturity 15 / 25
Community 5 / 25

How are scores calculated?

Stars

16

Forks

1

Language

Python

License

MIT

Last pushed

Oct 13, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/berkedenizbozyigit/novel-artificial-neurons"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.