Learning with Multiple Labels

By Rong Jin, Zoubin Ghahramani. Neural Information Processing Systems, 2002

408 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/89c3aed3e1219985c4a832f09adcd1df7096bf60

How researchers see this paper

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

Citations in the corpus

Ambiguity-aware Ensemble Training for Semi-supervised Dependency Parsing (2014, background): “Similar ideas of learning with ambiguous labelings are previously explored for classiﬁcation (Jin and Ghahramani, 2002) and sequence labeling problems (Dredze et al., 2009).”
A Study of the Robustness of KNN Classifiers Trained Using Soft Labels (2006, methodology): “In [10] an integrated framework for a generalized nearest prototype classifier (GNPC) is proposed.”
Ieee Transactions on Knowledge and Data Engineering 1 Multi-label Neural Networks with Applications to Functional Genomics and Text Categorization (unknown): “Note that although most works on multi-label learning assume that an instance can be associated with multiple valid labels, there are also works assuming that only one of the labels associated with an instance is correct [14] 1 .”
9 Entropy Regularization (methodology): “The optimization process itself is identical to the generalized EM described in section 9.3.1 with hard assignments [Grandvalet, 2002, Jin and Ghahramani, 2003].”
Multi-Instance Multi-Label Learning with Application to Scene Classification (2006, background): “(2)Although most works on multi-label learning assume that an instance can b e ssociated with multiple valid labels, there are also works assuming that only one of the labels associate d w th an instance is correct [6].”
Learning from ambiguously labeled images (2009, methodology): “We compare with the model proposed in [13], which is a discriminative model with an EM procedure adapted for the ambiguous label setting.”
Multi-Label Classification: An Overview (2007, background): “Jin and Ghahramani (2002) call multiple-label problems , the semi-supervised classification problems where each example is associated with more than one classes, but only one of those classes is the true class of the example.”
Large-Scale Object Classification Using Label Relation Graphs (2014, result): “By treating unobserved labels as latent variables, our approach also connects to prior work on learning from partial or incomplete labels [19,7,5].”
Multi-instance multi-label learning (2008, background): “The generality of multi-label problems, however, inevitably makes it more difficult to 1 Most work on multi-label learning assumes that an instance can be associated with multiple valid labels, but there is also some work assuming that only one of the labels among those associated with an instance i”
Reinforcement Learning Via Practice and Critique Advice (2010, unknown): “The multiple-label learning problem (Jin and Ghahramani 2003) interprets the label sets as possible labels only one of which is the true label.”
Neural Learning of One-of-Many Solutions for Combinatorial Problems in Structured Output Spaces (2020, methodology): “In PLL, the objective is to select the correct label out of many incorrect ones to reduce training noise, whereas in 1oML, selecting only one label for training provably improves the learnability and there is no question of reducing noise as all the labels are correct.”
Learning From Noisy Singly-labeled Data (2017, methodology): “…agreementbased algorithms jointly model worker skills and ground truth labels, estimating both using EM or similar techniques (Dawid and Skene, 1979; Jin and Ghahramani, 2003; Whitehill et al., 2009; Welinder et al., 2010; Zhou et al., 2012; Liu et al., 2012; Dalvi et al., 2013; Liu et al., 2012).”
Get another label? improving data quality and data mining using multiple, noisy labelers (2008, background): “A closely related problem setting is described by Jin and Ghahramani [10].”
A Review on Multi-Label Learning Algorithms (2014, background): “On one case, labels assigned to an training example may be false positive [81], [107].”