Adapting ImageNet-scale models to complex distribution shifts with self-learning

by   Evgenia Rusak, et al.

While self-learning methods are an important component in many recent domain adaptation techniques, they are not yet comprehensively evaluated on ImageNet-scale datasets common in robustness research. In extensive experiments on ResNet and EfficientNet models, we find that three components are crucial for increasing performance with self-learning: (i) using short update times between the teacher and the student network, (ii) fine-tuning only few affine parameters distributed across the network, and (iii) leveraging methods from robust classification to counteract the effect of label noise. We use these insights to obtain drastically improved state-of-the-art results on ImageNet-C (22.0 techniques yield further improvements in combination with previously proposed robustification methods. Self-learning is able to reduce the top-1 error to a point where no substantial further progress can be expected. We therefore re-purpose the dataset from the Visual Domain Adaptation Challenge 2019 and use a subset of it as a new robustness benchmark (ImageNet-D) which proves to be a more challenging dataset for all current state-of-the-art models (58.2 to guide future research efforts at the intersection of robustness and domain adaptation on ImageNet scale.


page 3

page 18


Pre-Training Transformers for Domain Adaptation

The Visual Domain Adaptation Challenge 2021 called for unsupervised doma...

Self-ensembling for visual domain adaptation

This paper explores the use of self-ensembling for visual domain adaptat...

Robust Ensembling Network for Unsupervised Domain Adaptation

Recently, in order to address the unsupervised domain adaptation (UDA) p...

Impact of ImageNet Model Selection on Domain Adaptation

Deep neural networks are widely used in image classification problems. H...

Tackling Partial Domain Adaptation with Self-Supervision

Domain adaptation approaches have shown promising results in reducing th...

Out of Distribution Detection on ImageNet-O

Out of distribution (OOD) detection is a crucial part of making machine ...

Are we done with ImageNet?

Yes, and no. We ask whether recent progress on the ImageNet classificati...