Latent Shift - A Simple Autoencoder Approach to Counterfactual Generation
The idea
Read the paper: https://arxiv.org/abs/2102.09475
Watch a video: https://www.youtube.com/watch?v=1fxSDP8DheI
The main diagram:
Animations/GIFs
| Smiling |
Arched Eyebrows |
|---|---|
 |
 |
| Mouth Slightly Open |
Young |
|---|---|
 |
 |
Generating a transition sequence
For a predicting of smiling
Multiple different targets
Comparison to traditional methods
For a predicting of pointy_nose
Getting Started
$pip install latentshiftimport latentshift
# Load classifier and autoencoder
model = latentshift.classifiers.FaceAttribute()
ae = latentshift.autoencoders.Transformer(weights="celeba")
# Load image
input = torch.randn(1, 3, 1024, 1024)
# Defining Latent Shift module
attr = captum.attr.LatentShift(model, ae)
# Computes counterfactual for class 3.
output = attr.attribute(input, target=3)
