This code helps to undertand the concept of Autoencoders. The autoencoder is trained on mnist digit dataset and it learns encoding of 64 units from an input of 784 pixels. This is a two step procedure.
- Encoder - which learns embedding from the input dimensions.
- Decoder - which recreates the image from the embedding created by the encoder.
Types of Autoencoders used
- Simple Network
- Deep Network
- Convolutional Network
You can install Conda for python which resolves all the dependencies for machine learning.
pip install requirements.txt
An autoencoder is a type of artificial neural network used to learn efficient data codings in an unsupervised manner. The aim of an autoencoder is to learn a representation (encoding) for a set of data, typically for the purpose of dimensionality reduction. Recently, the autoencoder concept has become more widely used for learning generative models of data.
We are using the mnist digit dataset.
- Network Used- Simple Network, Deep Network, Convolutional Network
- Technique - Autoencoders
If you face any problem, kindly raise an issue
- First, run
Coder.pywhich will train a simple, deep and a convolutional autoencoder and store it in h5 filr.
- Now you need to have the data, run
AutoencoderApp.pywhich will use computer vision to get the drawn on screen, encodes it and then decodes it to display the image.
- For altering the model, check
- For tensorboard visualization, go to the specific log directory and run this command
tensorboard --logdir=.You can go to
localhost:6006for visualizing your loss function.