Need help with Numpy-CNN?
Click the “chat” button below for chat support from the developer who created it, or find similar developers for support.

About the developer

Alescontrela
147 Stars 68 Forks 16 Commits 2 Opened issues

Description

A numpy-only implementation of a Convolutional Neural Network, from the ground up.

Services available

!
?

Need anything else?

Contributors list

Numpy-CNN

A numpy-only implementation of a Convolutional Neural Network, from the ground up.

Written by Alejandro Escontrela for this article on CNNs.

Purpose

To gain a quality understanding of convolutional neural networks and what makes them peform so well, I constructed one from scratch with NumPy. This CNN is in no way intended to replace popular DL frameworks such as Tensorflow or Torch, it is instead meant to serve as an instructional tool.

Training the network

To train the network on your machine, first install all necessary dependencies using:

$ pip install -r requirements.txt

Afterwards, you can train the network using the following command:

$ python3 train_cnn.py '.pkl'

Replace

 with whatever file name you would like. The terminal should display the following progress bar to indicate the training progress, as well as the cost for the current training batch:

portfolio_view

After the CNN has finished training, a .pkl file containing the network's parameters is saved to the directory where the script was run.

The network takes about 5 hours to train on my macbook pro. I included the trained params in the GitHub repo under the name

params.pkl
 . To use the pretrained params when measuring the network's performance, replace
 with params.pkl.

Measuring Performance

To measure the network's accuracy, run the following command in the terminal:

$ python3 measure_performance.py '.pkl'

This command will use the trained parameters to run predictions on all 10,000 digits in the test dataset. After all predictions are made, a value displaying the network's accuracy will appear in the command prompt:

portfolio_view

We use cookies. If you continue to browse the site, you agree to the use of cookies. For more information on our use of cookies please see our Privacy Policy.