MixStyle Neural Networks for Domain Generalization and Adaptation

07/05/2021
by   Kaiyang Zhou, et al.
0

Convolutional neural networks (CNNs) often have poor generalization performance under domain shift. One way to improve domain generalization is to collect diverse source data from multiple relevant domains so that a CNN model is allowed to learn more domain-invariant, and hence generalizable representations. In this work, we address domain generalization with MixStyle, a plug-and-play, parameter-free module that is simply inserted to shallow CNN layers and requires no modification to training objectives. Specifically, MixStyle probabilistically mixes feature statistics between instances. This idea is inspired by the observation that visual domains can often be characterized by image styles which are in turn encapsulated within instance-level feature statistics in shallow CNN layers. Therefore, inserting MixStyle modules in effect synthesizes novel domains albeit in an implicit way. MixStyle is not only simple and flexible, but also versatile – it can be used for problems whereby unlabeled images are available, such as semi-supervised domain generalization and unsupervised domain adaptation, with a simple extension to mix feature statistics between labeled and pseudo-labeled instances. We demonstrate through extensive experiments that MixStyle can significantly boost the out-of-distribution generalization performance across a wide range of tasks including object recognition, instance retrieval, and reinforcement learning.

READ FULL TEXT

page 5

page 6

research
04/05/2021

Domain Generalization with MixStyle

Though convolutional neural networks (CNNs) have demonstrated remarkable...
research
12/07/2020

Selective Pseudo-Labeling with Reinforcement Learning for Semi-Supervised Domain Adaptation

Recent domain adaptation methods have demonstrated impressive improvemen...
research
03/09/2022

Dynamic Instance Domain Adaptation

Most existing studies on unsupervised domain adaptation (UDA) assume tha...
research
12/07/2015

Scalable domain adaptation of convolutional neural networks

Convolutional neural networks (CNNs) tend to become a standard approach ...
research
07/25/2017

Improving Robustness of Feature Representations to Image Deformations using Powered Convolution in CNNs

In this work, we address the problem of improvement of robustness of fea...
research
07/25/2018

Two at Once: Enhancing Learning and Generalization Capacities via IBN-Net

Convolutional neural networks (CNNs) have achieved great successes in ma...
research
11/19/2021

Semi-Supervised Domain Generalization in Real World:New Benchmark and Strong Baseline

Conventional domain generalization aims to learn domain invariant repres...

Please sign up or login with your details

Forgot password? Click here to reset