Learning from Multiple Sources

By Koby Crammer, Michael Kearns, Jennifer Wortman Vaughan. 2006

375 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Representation & Self-Supervised Learning, Meta & Continual 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 Representation & Self-Supervised Learning

In Meta & Continual Learning

Citations in the corpus

Multiple-Source Adaptation with Domain Classifiers (2020, background): “Duan et al. [2009, 2012] further assumed plenty of unlabeled target data to form a meaningful regularizer, and a small set of labeled target data for training. Khosla et al. [2012], Blanchard et al.”
Algorithms and Theory for Multiple-Source Adaptation (2018, background): “Crammer et al. (2008) dealt with a problem with multiple sources distinct from domain adaptation where the sources have the same input distribution but can have different labels, modulo some disparity constraints. (See also discussion by Mansour et al. (2009a)). Gong et al. (2012) proposed a Rank of”
Adaptive Personalized Federated Learning (2020, background): “As pointed out, related works by Ben-David et al. (2010, 2007); Crammer et al. (2008) did the pioneering analysis of learning a single model from diﬀerent domains.”
Model Zoo: A Growing Brain That Learns Continually (2021, background): “To gain some intuition, we can connect this deﬁnition to a certain triangle inequality between the tasks developed by Crammer et al. (2008): in the realizable setting where e P i ( h ∗ i ) = 0 , for c, ρ ij = 1 , we can write (5) as e P i ( h ) + e P j ( h ∗ i ) ≥ e P j ( h ) which is akin to a…”
Moment Matching for Multi-Source Domain Adaptation (2018, background): “Originated from the early theoretical analysis [1, 28, 4], MSDA has many practical applications [45, 6].”
Multiple Source Adaptation and the Rényi Divergence (2009, background): “A problem with multiple sources but distinct from domain adaptation has also been considered by [5] where the sources have the same input distribution but can have different labels, modulo some disparity constraints.”
Class-Incremental Domain Adaptation (2020, methodology): “A line of work called Unsupervised Domain Adaptation (UDA) [2,3,5,21,25,31,22,51] has emerged that offers an elegant solution to the domain-shift problem.”
A Literature Survey on Algorithms for Multi-label Learning (2010, methodology): “There exists several methods in literature to address this problem of learning multiple tasks simultaneously [30, 10].”
A survey of multi-source domain adaptation (2015, background): “[27] also gave a bound for (noise-free) binary classification and (noise-free) regression.”
Structure compilation: trading structure for features (2008, background): “Theorem 1 generalizes Lemma 3 of Crammar et al. (2007) to conditional distributions and the case where p0 is not necessarily in an exponential family.”
Learning Multiple Tasks using Shared Hypotheses (2012, background): “[11, 10] whom developed theory for learning a model with few datasets from various tasks, assuming they are sampled from the same source.”
Domain Adaptation by Mixture of Alignments of Second-or Higher-Order Scatter Tensors (2016, background): “Single- or multiple-source [4] setups are possible e.”
Unsupervised Domain Adaptation with a Relaxed Covariate Shift Assumption (2017, methodology): “…are based on specific notions of distance between domains like the Adistance defined in Kifer, Ben-David, and Gehrke (2004) and used, most notably, in Ben-David et al. (2007), and like the discrepancy distance-based Rademacher complexity bound introduced in Mansour, Mohri, and Rostamizadeh (2009a).”
Domain Adaptation with Multiple Sources (2008, background): “Crammer et al. [5] [6] also addressed a problem where multiple sources are present but the nature of the problem differs from adaptation since the distribution of the input points is the same for all these sources, only the labels change due to varying amounts of noise.”
Generalization Bounds for Transfer Learning under Model Shift (2015, background): “Other existing work focuses on ﬁnding the set of sources that are closely re-lated to the target (Crammer et al. (2008)), or a reweighting of sources based on prediction errors (Yao and Doretto, (2010)).”