Generalizing from Several Related Classification Tasks to a New Unlabeled Sample

By Gilles Blanchard, Gyemin Lee, Clayton D. Scott. 2011

544 citations recorded by the source corpus.

Subfields: 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 Representation & Self-Supervised Learning

Citations in the corpus

Measuring Robustness to Natural Distribution Shifts in Image Classification (2020, background): “[6] Blanchard, G.”
Kernel Mean Embedding of Distributions: A Review and Beyonds (2016, methodology): “…(Sgouritsa et al. 2013, Chen et al. 2014, Schölkopf et al. 2015), covariate shift (Gretton et al. 2009, Pan et al. 2011) and domain adaptation (Blanchard et al. 2011, Muandet et al. 2013), selection bias correction (Huang et al. 2007), herding (Chen et al. 2010, Huszar and Duvenaud 2012),…”
Multiple-Source Adaptation with Domain Classifiers (2020, background): “[2012], Blanchard et al. [2011] combined all the source data to jointly train a single predictor.”
Algorithms and Theory for Multiple-Source Adaptation (2018, background): “Blanchard et al. (2011) presented MSA algorithms with theoretical guarantees. However, it combines all source data and target data to learn a final predictor. This paper also makes the strong assumption that the source and the target domains are i.i.d. realizations of some distribution, and their le”
Domain Generalization via Conditional Invariant Representations (2018, background): “In medical diagnosis, the distribution of cell types varies from patient to patient, and we need to train a classi-ﬁer on the data collected from previous patients that generalizes well to unseen patients (Blanchard, Lee, and Scott 2011; Muandet, Balduzzi, and Sch¨olkopf 2013).”
Multi-Source Domain Adaptation: A Causal View (2015, methodology): “…shift (Huang et al. 2007), the pooling SVM (denoted pool SVM ), which merge all source data to train the SVM, domain-invariant component analysis (DICA) (Muandet, Bal-duzzi, and Sch¨olkopf 2013), and Learning marginal predictors (LMP) proposed by (Blanchard, Lee, and Scott 2011) are also included.”
Domain Generalization via Invariant Feature Representation (2013, background): “In this case, the data exhibits covariate (or dataset) shift (Widmer and Kurat 1996, Quionero-Candela et al. 2009, Bickel et al. 2009b): although the marginal distributions P X on cell attributes vary due to biological or technical variations, the functional relationship P ( Y | X ) across diﬀerent…”
Learning from Distributions via Support Measure Machines (2012, methodology): “More recently, the empirical kernel (4) was employed in an unsupervised way for multi-task learning to g eneralize to a previously unseen task [16].”
Continual learning: a feature extraction formalization, an efficient algorithm, and fundamental obstructions (2022, background): “This has resulted in the proliferation of new concepts and settings such as out-of-distribution generalization [15], domain generalization [3], multi-task learning [39], continual learning [23] and etc.”
In Search of Lost Domain Generalization (2020, background): “The rest of this paper studies the problem of domain generalization, an extension of supervised learning where training datasets from multiple domains (or environments) are available to train our predictor [Blanchard et al., 2011].”
Unified Deep Supervised Domain Adaptation and Generalization (2017, background): “Specifically, [5] pulls all of the training data together in one dataset, and learns a single SVM classifier.”
WILDS: A Benchmark of in-the-Wild Distribution Shifts (2020, neutral): “…(Saenko et al., 2010; Gong et al., 2012), sentiment analysis (Blitzer et al., 2007; Pan et al., 2010), part-of-speech tagging (Marcus et al., 1993; Daumé III, 2007), flow cytometry (Blanchard et al., 2011; Muandet et al., 2013), or land cover classification (Bruzzone and Marconcini, 2009).”
Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations (2022, methodology): “Related methods have been developed in several areas of machine learning, such as ML Fairness [22, 30, 47, 77, 2, 46], domain adaptation [23, 24] and domain generalization [9, 65, 57, 29, 84] including works on Invariant Risk Minimization and causality [75, 4, 53, 5].”
Scatter Component Analysis: A Unified Framework for Domain Adaptation and Domain Generalization (2015, background): “[11] first studied this issue and proposed an augmented SVM that encodes empirical marginal distributions into the kernel for solving automatic gating of flow cytometry.”
A Survey of Unsupervised Deep Domain Adaptation (2018, unknown): “eature learning approaches similar to those discussed in Section3.1have been used for the related problem of domain generalization, where there are multiple source domains and an unseen target domain [17, 173]. Zhao et al. [284] use an adversarial approach with a domain classifier to learn a model o”
Multi-task Learning with Labeled and Unlabeled Tasks (2016, background): “Note, also, that a similar assumption appears implicitly in [9].”
Adaptive Risk Minimization: A Meta-Learning Approach for Tackling Group Shift (2020, methodology): “This assumption is used by, e.g., group distributionally robust optimization (DRO) [25, 55], robust federated learning [46, 34], and domain generalization [4, 18].”