Overview • About us • Features • Install • Usage • Credits • License • Donate
Overview
MNIST is a dataset of handwritten digits, widely used for training and testing image classification models, and DNN stands for Deep Neural Network, a type of artificial neural network that has multiple layers of nodes.
MNIST DNN is web application that allows users to create and configure their own DNN models using MNIST dataset.
The project consists of two packages:
MNIST_DNN
Provides features to create and configure MNIST DNN models.
MNIST_DNN_API
API service package that uses MNIST_DNN to create an API that retrieves user request to create specific models and return the trained models back to the user.
With these packages, we can create our own web application that provides user friendly interfaces to play with MNIST dataset.
About us
The MNIST DNN Project was started to give us the opportinuity to create our own neural network model without using any deep learning APIs like Keras.
Ultimately, our goal is to gain a deeper understanding of machine learning and to gain experience with various techniques, not just those related to machine learning. (i.e. REST API, Web application, package deployment, etc.)
John Yoon
University of Waterloo | Computer Science 3B Co-op
Features
Features are still in development phase, please check out our wiki for more detail.
Install
Product is still in development phase, please check out our wiki for more detail.
Usage
Product is still in development phase, please check out our wiki for more detail.
Credits
Special thanks
The project uses following packages:
- numpy
- pandas
- matplotlib
- flask
License
Donate
