Deeper, Broader and Artier Domain Generalization

By Da Li, Yongxin Yang, Yi-Zhe Song, Timothy M. Hospedales. 2017

1,792 citations recorded by the source corpus.

The problem of domain generalization is to learn from multiple training domains, and extract a domain-agnostic model that can then be applied to an unseen domain. Domain generalization (DG) has a clear motivation in contexts where there are target domains with distinct characteristics, yet sparse data for training. For example recognition in sketch images, which are distinctly more abstract and rarer than photos. Nevertheless, DG methods have primarily been evaluated on photo-only benchmarks focusing on alleviating the dataset bias where both problems of domain distinctiveness and data sparsity can be minimal. We argue that these benchmarks are overly straightforward, and show that simple deep learning baselines perform surprisingly well on them. In this paper, we make two main contributions: Firstly, we build upon the favorable domain shift-robust properties of deep learning methods, and develop a low-rank parameterized CNN model for end-to-end DG learning. Secondly, we develop a DG benchmark dataset covering photo, sketch, cartoon and painting domains. This is both more practically relevant, and harder (bigger domain shift) than existing benchmarks. The results show that our method outperforms existing DG alternatives, and our dataset provides a more significant DG challenge to drive future research.

Subfields: Artificial Intelligence, Cancer Research, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Cancer Research

In Computer Vision and Pattern Recognition

In Representation & Self-Supervised Learning

Citations in the corpus

Shortcut learning in deep neural networks (2020, neutral): “otation and viewpoint [98]. PACS and other domain generalisation datasets require extrapolation beyond i.i.d. data per design by testing on a domain different from training data (e.g. cartoon images) [99]. Shift-MNIST / biased CelebA / unfair dSprites are controlled toy datasets that introduce corre”
Learning Robust Global Representations by Penalizing Local Predictive Power (2019, methodology): “We test on the PACS data set (Li et al., 2017a), which consists of collections of images over four domains, including photo, art painting, cartoon, and sketch.”
High Fidelity Visualization of What Your Self-Supervised Representation Knows About (2021, methodology): “We used cell images from microscope and a photo of a status (Both from Wikimedia Commons), sketch and cartoons from PACS (Li et al., 2017), image segmentation from Cityscape (Cordts et al.”
Learning Robust Representations by Projecting Superficial Statistics Out (2018, methodology): “Finally, we tested on the PACS data set (Li et al., 2017a), which consists of collections of images of seven different objects over four domains, including photo, art painting, cartoon, and sketch.”
SWAD: Domain Generalization by Seeking Flat Minima (2021, background): “For example, PACS dataset [7] has seven task labels ( e.g. , “dog”, “horse”) and four domain labels ( e.g. , “photo”, “sketch”).”
Moment Matching for Multi-Source Domain Adaptation (2018, neutral): “Dataset Year Images Classes Domains Description Digit-Five - ∼100,000 10 5 digit Office [37] 2010 4,110 31 3 office Office-Caltech [11] 2012 2,533 10 4 office CAD-Pascal [33] 2015 12,000 20 6 animal,vehicle Office-Home [43] 2017 15,500 65 4 office, home PACS [21] 2017 9,991 7 4 animal, stuff Open MI”
Learning to Generalize: Meta-Learning for Domain Generalization (2017, positive): “The few existing DG methods typically train on multiple source domains and propose mechanisms to extract some domain agnostic representation or model that describes common aspects of known domains (Khosla et al. 2012; Muandet, Balduzzi, and Schölkopf 2013; Ghifary et al. 2015; Li et al. 2017).”
In Search of Lost Domain Generalization (2020, methodology): “Dataset Architecture Colored MNIST MNIST ConvNetRotated MNIST PACS ResNet-50 VLCS Office-Home TerraIncognita DomainNet Table 7: Details of our MNIST ConvNet architecture.”
Meta-Learning in Neural Networks: A Survey (2020, neutral): “PACS [262] provides a good starter benchmark, with Visual Decathlon [42], [221] and MetaDataset [109] providing larger scale alternatives.”
WILDS: A Benchmark of in-the-Wild Distribution Shifts (2020, neutral): “It is also common to split the data to induce a shift that is more extreme than one might typically encounter, e.g., in PACS (Li et al., 2017), which includes classifying photos solely from cartoons and other stylized representations; DeepFashion Remixed (Hendrycks et al., 2020), which involves…”
Generalizing to Unseen Domains via Adversarial Data Augmentation (2018, background): “Domain generalization methods [28, 22, 27, 33, 24] that propose different ways to better generalize to unknown domains are also related to our work.”
Domain Adaptive Faster R-CNN for Object Detection in the Wild (2018, background): “Recent works aim to improve the domain adaptability of deep neural networks, including [40, 15, 18, 50, 45, 43, 34, 24, 41, 42].”
Non-stationary Transformers: Exploring the Stationarity in Time Series Forecasting (2022, background): “Besides, it is a fundamental problem to make deep models generalize well on a varying distribution [28, 21, 5].”
Deep Domain-Adversarial Image Generation for Domain Generalisation (2020, methodology): “To evaluate DDAIG, we conduct extensive experiments on three DG benchmark datasets, namely PACS (Li et al. 2017), Ofﬁce-Home (Venkateswara et al. 2017) and digit recognition among MNIST (LeCun et al. 1998), MNIST-M (Ganin and Lempitsky 2015), SVHN (Netzer et al. 2011) and SYN (Ganin and Lempitsky 20”
Out-of-Distribution Generalization via Risk Extrapolation (REx) (2020, methodology): “Domain Generalization: VLCS and PACS In this section we compare REx, IRM and ERM performance on the VLCS (Torralba & Efros, 2011) and PACS (Li et al., 2017) image datasets.”
Agnostic Domain Generalization (2018, methodology): “Still, there has been recently a growing interest on how to leverage over multiple source domains when unlabeled target data are available [47], and even more on how to generalize over any possible target domain, when it is not possible to access target data of any sort a priori [23].”