https://github.com/vasukalariya/neural-network-from-scratch

It's a simple Feedforward Neural Network implemented just using Numpy
https://github.com/vasukalariya/neural-network-from-scratch

artificial-neural-networks classification mnist neural-network numpy

Last synced: 3 months ago
JSON representation

It's a simple Feedforward Neural Network implemented just using Numpy

• Host: GitHub
• URL: https://github.com/vasukalariya/neural-network-from-scratch
• Owner: vasukalariya
• License: apache-2.0
• Created: 2019-06-28T17:53:56.000Z (about 7 years ago)
• Default Branch: master
• Last Pushed: 2019-09-03T17:44:12.000Z (almost 7 years ago)
• Last Synced: 2026-01-03T01:05:21.251Z (7 months ago)
• Topics: artificial-neural-networks, classification, mnist, neural-network, numpy
• Language: Python
• Homepage:
• Size: 15.6 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Neural-Network-from-Scratch

## It's a simple Feedforward Neural Network implemented just using Numpy library. :smiley:

* Libraries:

  * Numpy: Matrix calculation and most of the code !!!.

  * Tensorflow: Dataset (MNIST).

  * Matplotlib: Sample Image Display.

* Files:

  * archi.py: Contains the basic architecture of Layer(Dense) and Model.

  * test.py: Contains the sample test used on MNIST dataset.

* Note:

  * You can change the architecture of the model other than that used in test.py.:thumbsup:

  * Also can use a different datatset but remember the dimensions.

    * x_train: (no_examples, features)

    * y_train: (no_examples, target)

* Architecture of the model must be a list of dictionary containing values of no of units in the layer and activation function.

* test.py already has converted one-hot encoding for MNIST you may use your knowledge and change it according to your dataset.