A neural network in python for MNIST Handwritten digit database.
The Network is based on a basic 3 layer architecture using only the built in functionalities and libraries of python.
Repository Link : https://web.archive.org/web/20150906081542/http://yann.lecun.com/exdb/mnist/index.html
The dataset is first converted into csv format before loading in the network.
Performance Results:
- Total number of correct predictions 9306.0
- Total predictions 10000.0
- performance = 0.9306
Future work
- Improve the accuracy.
- Create a Learning rate vs accuracy graph.
- Move the functionality to scipy to reduce processing time