vector-field-based-neural-network

This is a repo for re-create the experiment is paper: vector field based neural networks. Link: https://arxiv.org/abs/1802.08235

Main steps:

1. create the sin, moons and circle datasets
2. create the vector field layers
3. construct the vector field based network
4. plot vector field
5. plot transformed space
6. plot space distortion
7. check the result
8. work with iris dataset

cli-cmds:

git clone https://github.com/gitathrun/vector-field-based-neural-network.git cd vector-field-based-neural-network

data is already in data directory figures are in figure directory if you want to generate the data by yourself, try the cmd make sure you are in python 3.5 envionment python artificial_data_generate.py

Useful material for figure drawing:

1. vector field: https://matplotlib.org/api/_as_gen/matplotlib.pyplot.quiver.html https://matplotlib.org/examples/pylab_examples/quiver_demo.html https://scipython.com/blog/visualizing-a-vector-field-with-matplotlib/
2. space distortion: https://stackoverflow.com/questions/47295473/how-to-plot-using-matplotlib-python-colahs-deformed-grid