https://medium.com/@mehmet.ozel2701/entangled-ai-how-neural-networks-can-learn-from-each-other-without-sharing-data-beeeeadba451

This project demonstrates an experimental framework for entangled learning between two heterogeneous models (a Convolutional Neural Network and a Multi-Layer Perceptron) using the MNIST dataset.

• Models: CNN and MLP
• Dataset: MNIST
• Loss Function: Categorical Crossentropy + KL Divergence (Entangled Loss)
• Entanglement Strength (λ): Increases dynamically over epochs

• entangled_models_final.ipynb - Full Jupyter notebook with training and results
• entangled_utils.py - Helper functions for entangled loss and lambda scheduler
• requirements.txt - Dependencies

• CNN Accuracy: ~99.6%
• MLP Accuracy: ~98.7%
• Demonstrates effective information transfer via entangled output feedback

MIT