neural-networks

Neural networks written in numpy. Current NNs:

1. 2-layer FC (basic)
2. 3-layer FC

• Classifies whether an image is a cat or not.
• Depends on: datasets/train_catvnoncat.h5, datasets/test_catvnoncat.h5, load_dataset

3. RNN

• Character-level RNN based off @karpathy's gist (https://gist.github.com/karpathy/d4dee566867f8291f086)
• Adagrad Optimizer
• Depends on: datasets/shakespearean.txt. Can be used with datasets/hp1.txt

4. LSTM

• Character-level shallow LSTM
• Adam Optimizer, Gradient Checking
• Depends on: datasets/shakespearean.txt. Can be used with datasets/hp1.txt

Please let me know if you have any questions.

The scripts were written with the purpose of informing my understanding of NNs on a fundamental level. While I have tried to keep the code clean, my priority did not lie in writing clever Python code.

BT