https://github.com/ritzvik/simpleneuralnetwork
https://github.com/ritzvik/simpleneuralnetwork
deep-learning dense-neural-network least-square-regression minimization neural-network numpy python sigmoid-function
Last synced: about 2 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/ritzvik/simpleneuralnetwork
- Owner: ritzvik
- Created: 2018-10-22T15:14:12.000Z (over 7 years ago)
- Default Branch: master
- Last Pushed: 2018-10-22T17:16:20.000Z (over 7 years ago)
- Last Synced: 2024-12-29T12:43:30.508Z (about 1 year ago)
- Topics: deep-learning, dense-neural-network, least-square-regression, minimization, neural-network, numpy, python, sigmoid-function
- Language: Python
- Size: 18.3 MB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# SimpleNeuralNetwork
## Dense Neural Network Implementation using only Numpy
### hdigits.npz
numpy.save_compressed version of MNIST handwriiten digits dataset. Usage :
```
npzfile = np.load('hdigits.npz')
train_images = npzfile['mnist_train_images']
train_labels = npzfile['mnist_train_labels']
test_images = npzfile['mnist_test_images']
test_labels = npzfile['mnist_test_labels']
```
arrays would be of shape (n,1,28,28) where n denotes number of instances, 1 is number of channels in images, and 28 * 28 is dimensions for train and test images
labels are 1D numpy array
### nnClass.py
Contains the class defination of dense neural network having sigmoid activation function.
Can be directly used or used to make components of bigger, more complex networks
### train.py
Trains on MNIST train dataset and saves the network using pickle.
### test.py
Tests MNIST test dataset on the model made by train.py and prints accuracy.