Skip to content

Convolutional neural network implemented in pytorch achieving a 99.71% test accuracy on the EMNIST dataset

Notifications You must be signed in to change notification settings

austin-hill/EMNIST-CNN

Repository files navigation

EMNIST-Convolutional-Neural-Net

A convolutional neural network implemented in PyTorch that achieved a 99.71% classification accurracy on the EMNIST dataset of digits after 60 epochs of training, without using an ensemble of networks. I included a checkpoint at the 99.71% accurracy.

Requirements

pip install emnist matplotlib torch==1.9.0+cu111

Sources

I chose to use two convolutional layers and two fully connected layers based on results from https://www.kaggle.com/cdeotte/how-to-choose-cnn-architecture-mnist. The rest of the parameters I chose myself using the plots generated by the train_plot_params function. Credit for the EMNIST dataset goes to: Cohen, G., Afshar, S., Tapson, J., & van Schaik, A. (2017). EMNIST: an extension of MNIST to handwritten letters. Retrieved from http://arxiv.org/abs/1702.05373.

About

Convolutional neural network implemented in pytorch achieving a 99.71% test accuracy on the EMNIST dataset

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages