Simple-nn: Simple neural network implemented using numpy+scipy
These are practices I made to help gain a deeper understanding of the concepts of inner workings of neural networks. They are certainly not as efficient and robust as those mature implementations like Tensorflow, keras, and only suitable for educational purposes.
Types of networks
Each network type will have:
- One or more pieces of articles explaining the workings of the network.
- A Python implementation using
numpy
+scipy
. - An example application of the network.
1. A 5-layer multilayer perceptron network for hand-written digit classification
2. A 5-layer convolutional neural network for hand-written digit classification
more to come
Contribution
These are my personal understandings of the subject and therefore are most likely to have mistakes. If you spot any of such, please fire an issue and let me know.