This repository contains codes illustrating how deep deducing plays Game of Life. To reproduce the results in the paper, simply run Deducing_xxx.py
Packages requirements:
numpy
scipy
To train sets of weight matrix in the paper, run this command:
Learning.py
To use sets of trained weight matrix to have each cell in the present game table to deduce its legit status in the next generation, run this command:
Deducing_xxx.py
We have sets of weight matrix already-trained for your convenience. Please see the definition below to understand the content of each set of weight matrix.
self.Conway _ 1 _ 100x100x100 _ 30 _ 0.000001 _ 50m _ [1] _ weight_list
means a single set of weight matrix.
The definition for each bracket (from left to right) is listed below:
Conway
means this neural network is trained to mimic the rationale of the cells in Conway's Game of Life.
1
means the sample batch in the learning phase for this trained neural network is 1 per each learning epoch.
100x100x100
means the trained neural network has three hidden layers, each with 100 neurons.
30
means the initial value for the set of slope multiplie to be updated in the learning phase.
0.000001
means the learning rate in the learning phase for this trained neural network is 0.000001.
50m
means the learning epochs in the learning phase for this trained neural network is 50 million or 5*10^7. The learning epochs are usually big in order to force the neural network to over-fit.
[1]
means the label of this trained neural network under the above training condition.
For example, if it is [3], then it means this neural network is the third neural network under the training condition
Conway _ 1 _ 100x100x100 _ 30 _ 0.000001 _ 200m.
weight_list
means the set of weight matrix of this trained neural network.
For results, please refer to the figures in the paper.