https://github.com/soumyac1999/keras-mnist-nn
A 3-layer neural network trained on MNIST implemented on Keras
https://github.com/soumyac1999/keras-mnist-nn
Last synced: 4 months ago
JSON representation
A 3-layer neural network trained on MNIST implemented on Keras
- Host: GitHub
- URL: https://github.com/soumyac1999/keras-mnist-nn
- Owner: soumyac1999
- Created: 2018-07-07T16:39:28.000Z (almost 7 years ago)
- Default Branch: master
- Last Pushed: 2018-09-18T19:53:22.000Z (over 6 years ago)
- Last Synced: 2024-12-30T21:41:42.816Z (5 months ago)
- Language: Jupyter Notebook
- Homepage:
- Size: 1.37 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# keras-mnist-nn
A 3-layer neural network for MNIST using keras
The jupyter notebook contains my submission for the [Digit Recognizer Challenge](https://www.kaggle.com/c/digit-recognizer/) on Kaggle the same neural network. The neural network got a score of 0.97071 on 1 and the submission was ranked 1722 out of 2459 on a rolling scoreboard.
## Layers
- Input layer
- Layer with `392` neurons [relu activation]
- Layer with `196` neurons [relu activation]
- Output layer with `10` neurons [softmax activation]
This model achieves 98.5% test accuracy in 40 epochs using `rmsprop` optimizer
### To Use
Download [MNIST](http://yann.lecun.com/exdb/mnist/) dataset in a folder and name it 'mnist'