Learning from labeled and unlabeled data with label propagation

By Xiaojin Zhu, Zoubin Ghahramani. 2002

1,934 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

Data-Efficient Image Recognition with Contrastive Predictive Coding (2019, background): “t it makes predictions across space, but differs in that it also predicts representations across layers in the model. 4 A common alternative approach for improving data efﬁciency is label-propagation [65], where a classiﬁer is trained on a subset of labeled data, then used to label parts of the unla”
Semi-Supervised QA with Generative Domain-Adaptive Nets (2017, background): “The problem is challenging because conventional manifold-based semi-supervised learning algorithms (Zhu and Ghahramani, 2002; Yang et al., 2016a) cannot be straightforwardly applied.”
Virtual Adversarial Training: A Regularization Method for Supervised and Semi-Supervised Learning (2017, background): “same cluster, which is what we often desire in semi-supervised learning. As we show in Fig. 1, VAT on semi-supervised learning can be given a similar interpretation as that given to label propagation [44], which is another branch of methods for semi-supervised learning. 3.3 Fast Approximation Method”
Temporal Ensembling for Semi-Supervised Learning (2016, neutral): “owsky [25] in the context of linguistic analysis. Whitney and Sarkar [24] analyze Yarowsky’s algorithm and propose a novel graph-based label propagation approach. Similarly, label propagation methods [27] infer labels for unlabeled training data by comparing the associated inputs to labeled training”
Regularization With Stochastic Transformations and Perturbations for Deep Semi-Supervised Learning (2016, background): “For example, there are numerous online resources for images and video sequences of different types.”
Semi-supervised learning, causality, and the conditional cluster assumption (2019, background): “These construct a similarity-based graph representation of the data and propagate labels to neighbours in this graph [13, 14, 15].”
Attention-based Graph Neural Network for Semi-supervised Learning (2018, background): “There are signiﬁcant gaps in accuracy for those approaches not using the graph, i.e. T-SVM, and also those that use the graph differently, such as Label Propagation (LP) and Planetoid.”
Weight-averaged consistency targets improve semi-supervised deep learning results (2017, background): “Label propagation takes advantage of this assumption by pushing label information from each example to examples that are near it according to some metric (Zhu & Ghahramani, 2002).”
Smooth Neighbors on Teacher Graphs for Semi-Supervised Learning (2017, background): “Label propagation pushes label information [40] from an labeled instance to its neighbors using a predeﬁned distance metric.”
Bootstrapped Graph Diffusions (2018, methodology): “In particular, we discuss two textbook label propagations methods [42, 43] and a distance-diffusion method [11].”
SemiBoost: Boosting for Semi-Supervised Learning (2009, methodology): “The well known approaches in this category include Harmonic Function based approach [18], Spectral Graph Transducer (SGT) [7], Gaussian process base d approach [23], Manifold Regularization [12] and Label Prop agation approach [3], [4].”
MovieNet: A Holistic Dataset for Movie Understanding (2020, methodology): “81 body+face LP [90] 8.”
Semi-Supervised and Unsupervised Deep Visual Learning: A Survey (2022, methodology): “The idea shared by most existing works is to assign each unlabeled sample with a class label based on a certain underlying data structure, e.g., manifoldstructure[41],[43],andgraphstructure[45].”
Semi-supervised Learning on Graphs with Generative Adversarial Nets (2018, methodology): “ere is a long line of works about semi-supervised learning over graphs. One important category of the research is mainly based on the graph Laplacian regularization framework. For example, Zhu et al. [41] proposed a method called Label Propagation for learning from labeled and unlabeled data on grap”
Twitter Polarity Classification with Label Propagation over Lexical Links and the Follower Graph (2011, methodology): “Graph-based methods such as label propagation (Zhu and Ghahramani, 2002; Baluja et al., 2008; Talukdar and Crammer, 2009) provide a natural means to represent and exploit such relationships in order to improve classification, often while requiring less supervision than with standard classification.”