Deep Domain Confusion: Maximizing for Domain Invariance

TL;DR

This paper introduces a CNN architecture with an adaptation layer and domain confusion loss to learn domain-invariant features, improving performance on visual domain adaptation tasks without extensive fine-tuning.

Contribution

A novel CNN design incorporating a domain confusion loss and adaptation layer for effective domain invariance in visual tasks.

Findings

Empirical performance surpasses previous methods on benchmark tasks.

Domain confusion metric aids in model selection and layer positioning.

The approach reduces dataset bias in deep CNN models.

Abstract

Recent reports suggest that a generic supervised deep CNN model trained on a large-scale dataset reduces, but does not remove, dataset bias on a standard benchmark. Fine-tuning deep models in a new domain can require a significant amount of data, which for many applications is simply not available. We propose a new CNN architecture which introduces an adaptation layer and an additional domain confusion loss, to learn a representation that is both semantically meaningful and domain invariant. We additionally show that a domain confusion metric can be used for model selection to determine the dimension of an adaptation layer and the best position for the layer in the CNN architecture. Our proposed adaptation method offers empirical performance which exceeds previously published results on a standard benchmark visual domain adaptation task.