B657_Corum

B657 Final Project for Andrew Corum

To test this code, you will need TensorFlow installed on your machine. GPU acceleration (ie CuDNN) is recommended.

From the top directory, run python3 main.py to train and test a few pre-defind models (3 different RandWire models, AlexNet, TinyCNN, and HandmadeCNN).

To generate new random graphs, run python3 Graphs/graphs.py. The new graphs should be displayed in a window and saved to the Graphs/SavedGraphs/ directory.

General code-related references

• TensorFlow Docs: https://www.tensorflow.org/
• CUDA install/setup docs: https://docs.nvidia.com/cuda/cuda-installation-guide-microsoft-windows/
• Simple CNN for MNIST dataset: https://linux-blog.anracom.com/2020/05/31/a-simple-cnn-for-the-mnist-datasets-ii-building-the-cnn-with-keras-and-a-first-test/
• Code for AlexNet: https://towardsdatascience.com/alexnet-8b05c5eb88d4
• PyTorch implementation of RandWire: https://github.com/seungwonpark/RandWireNN/tree/0850008e9204cef5fcb1fe508d4c99576b37f995

Additional references can be found in the final report and specifically mentioned throughout the code.