Welcome to DomId’s documentation!

About DomID

The Variational Deep Embedding (VaDE) model is trained to learn lower-dimensional representations of images based on a Mixture-of-Gaussians latent space prior distribution while optimizing cluster assignments. In this package, examples on multiple dataset has been presented.

about_link

DomainLab

DomainLab is a submodule that has been used to develop DomID, and it aims at learning domain invariant features by utilizing data from multiple domains so the learned feature can generalize to new unseen domains.

Loading a Datasets and Defining a Task

Running the code with the custom dataset entails initialization of a Task and a Dataset. Examples of each for both HER2 and MNIST are included. More could be found in the sections below.

Datasets and Tasks

• Dataset Initialization Submodule
• Task Initialization Submodule

Composition and Defining a Model

Model is built from the building blocks in domid/compos directory. However, the model for the experiment is defined in the domid/models. For more details, see below.

Models

• Model Building Structures Submodule
  • Building blocks
  • Convolutional VAE
  • Linear VAE
  • Base MNIST Model
  • Predict Function
  • Store Results
  • Tensorboard Function
  • Module contents
• Model Initialization Module
  • M2YD Model
  • VaDE Model
  • Module contents

Training a Model

Training of the model consists of Observer and Trainer. .. toctree:

:maxdepth: 2
:caption: Training a Model

domid.algos
domid.trainers

Indices and tables

• Index

• Module Index

• Search Page