View Code? Open in Web Editor
NEW
This project forked from 1uc1f3r616/neurex
Neural Encryption Networks.
License: GNU Affero General Public License v3.0
Jupyter Notebook 97.80%
Python 2.20%
neurex's Introduction
![Binder](https://camo.githubusercontent.com/e91e1d353a8b6acf0b42547ac3901f2c30138a3abaaa3d3c242da30b5b4f8426/68747470733a2f2f6d7962696e6465722e6f72672f62616467655f6c6f676f2e737667)
- Uses fast auto-encoders that can encrypt the text.
- We create multiple hashmaps of varying bit length using HashMapGenerator.py.
- Hashmap creation is random. Could be tied to some markov chain as well.
- We learn that hashmap through our autoencoders.
- These autoencoders work simultaneously to provide an encryption. (encrpyter_multiple_loading_running.py file)
- Note multiple autoencders can be generated from hashmaps with different bit length settings. All can be combined.
- Takes < 2mins to create a new network.
- Little slower than normal encryption algos.
- Network V1.01 works for 6 bits for lowercase
- Network V1.11 works for for 8 bits lowercase
- Network V1.2.0 works for 8 bists all characters
- Network V1.2.1 works for 32 bits all characters
- Network V1.2.3 works for 56 bits all characters
- Parallel multiple neural network encryption algorithm aded.
- Not chained to processors / GPU yet.
neurex's People