Learning to Linearize Under Uncertainty

TL;DR

This paper introduces a new architecture and loss function for training deep feature hierarchies in an unsupervised manner by linearizing transformations in natural videos, incorporating latent variables to handle uncertainty.

Contribution

It proposes a novel generative model architecture with a specialized loss for unsupervised learning of deep hierarchies from video data, addressing uncertainty with latent variables.

Findings

Successfully linearizes transformations in video sequences

Improves unsupervised training of deep hierarchies

Handles uncertainty through latent variables

Abstract

Training deep feature hierarchies to solve supervised learning tasks has achieved state of the art performance on many problems in computer vision. However, a principled way in which to train such hierarchies in the unsupervised setting has remained elusive. In this work we suggest a new architecture and loss for training deep feature hierarchies that linearize the transformations observed in unlabeled natural video sequences. This is done by training a generative model to predict video frames. We also address the problem of inherent uncertainty in prediction by introducing latent variables that are non-deterministic functions of the input into the network architecture.