Best Sources Forward: Domain Generalization through Source-Specific Nets

By Massimiliano Mancini, Samuel Rota Bulò, Barbara Caputo, Elisa Ricci. 2018

173 citations recorded by the source corpus.

A long standing problem in visual object categorization is the ability of algorithms to generalize across different testing conditions. The problem has been formalized as a covariate shift among the probability distributions generating the training data (source) and the test data (target) and several domain adaptation methods have been proposed to address this issue. While these approaches have considered the single source-single target scenario, it is plausible to have multiple sources and require adaptation to any possible target domain. This last scenario, named Domain Generalization (DG), is the focus of our work. Differently from previous DG methods which learn domain invariant representations from source data, we design a deep network with multiple domain-specific classifiers, each associated to a source domain. At test time we estimate the probabilities that a target sample belongs to each source domain and exploit them to optimally fuse the classifiers predictions. To further improve the generalization ability of our model, we also introduced a domain agnostic component supporting the final classifier. Experiments on two public benchmarks demonstrate the power of our approach.

Subfields: Artificial Intelligence, 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 Computer Vision and Pattern Recognition

In Representation & Self-Supervised Learning

Citations in the corpus

Learning Robust Global Representations by Penalizing Local Predictive Power (2019, background): “…adversarial (or similar) techniques (Ghifary et al., 2015; Wang et al., 2016; Motiian et al., 2017b; Li et al., 2018a; Carlucci et al., 2018), or build ensembles of per-domain models that are then fused representations together (Bousmalis et al., 2016; Ding and Fu, 2018; Mancini et al., 2018).”
Learning Robust Representations by Projecting Superficial Statistics Out (2018, methodology): “We compared the performance of HEX/ADV with several methods tested on this data including CAE (Rifai et al., 2011), MTAE (Ghifary et al., 2015), CCSA (Motiian et al., 2017), DANN (Ganin et al., 2016), Fusion (Mancini et al., 2018), LabelGrad, and CrossGrad (Shankar et al., 2018).”
In Search of Lost Domain Generalization (2020, background): “Group Distributionally Robust Optimization (DRO, Sagawa et al. (2019)) performs ERM while increasing the importance of domains with larger errors. 3. Inter-domain Mixup (Mixup, Xu et al. (2019); Yan et al. (2020); Wang et al. (2020b)) performs ERM on linear interpolations of examples from random pai”
Learning to Optimize Domain Specific Normalization for Domain Generalization (2019, background): “Domain-speciﬁc information is handled by designated modules within deep neural networks [4, 9] while [14] proposes a soft model selection technique to obtain generalized representations.”
FSDR: Frequency Space Domain Randomization for Domain Generalization (2021, methodology): “Most existing DG methods can be broadly categorized into multi-source DG [44, 13, 12, 34, 16, 33, 43, 42, 59, 36, 2, 10, 35] and single-source DG [64, 70, 77, 50] both of which strive to learn domainagnostic features by either learning a domain-invariant feature space [44, 13, 12, 16, 34, 59, 36, 10”
Episodic Training for Domain Generalization (2019, methodology): “7% improvement on prior state-of-the-art methods MLDG [18], Fusion [23] and MetaReg [1].”
Deep Stable Learning for Out-Of-Distribution Generalization (2021, background): “A common approach is to extract domain-invariant features over multiple source domains [17, 25, 32, 34, 40, 10, 22, 43, 48, 40] or to aggregate domain-specific modules [36, 37].”
Domain Generalization via Gradient Surgery (2021, methodology): “Proposed works have mainly focused on: i) training and fusing multiple domain-specific models [33, 18], ii) learning and extracting common knowledge from multiple source domains, such as domain-invariant representations [20, 7, 16] or domainagnostic models [11, 15, 14], and iii) increasing the data ”
Generalizing to Unseen Domains: A Survey on Domain Generalization (2021, methodology): “[104] proposed to use learnable weights for aggregating the predictions from different source specific classifiers, where a domain predictor is used to predict the prob-”
Learning to Learn Single Domain Generalization (2020, methodology): “Most previous methods either tried to learn a domain-invariant space to align domains [32, 9, 12, 21, 59] or aggregate domain-speciﬁc modules [29, 28].”
Learning to Balance Specificity and Invariance for In and Out of Domain Generalization (2020, unknown): “40 Fusion [35] 64.”
Domain Generalization: A Survey (2021, methodology): “Single Pretext Task In addition to using the standard classiﬁcation loss, Carlucci et al. [49] taught a neural network to solve the Jigsaw puzzles problem [240], hoping that the network can learn regularities that are more generalizable across domains.”