This repository contains code which evaluates the abilities of Real, Complex, and quaternion based Dense Layers to solve the 4-bit parity check problem. It demonstrates the effectiveness of using other algebras to reduce the size of Neural Networks for a given task.
Install requirements for experiments with pip:
pip install numpy Theano keras tensorflow
Depending on your Python installation you might want to use anaconda or other tools.
- Run models:
python xor.py
Parameters can be modified directly in xor.py, such as:
- Number of epochs
- Batch Size
- Optimizer
This work is based on the original paper of Tohru Nitta: "A Solution to the 4-bit Parity Problem with a Single Quaternary Neuron"