Semi-Supervised Learning

By Sugato Basu. 2019

4,362 citations recorded by the source corpus.

Subfields: Artificial Intelligence, 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 Representation & Self-Supervised Learning

Citations in the corpus

A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark (2019, background): “…learning is studied in many contexts: supervised pre-training (Sharif Razavian et al., 2014), self-supervised learning (Doersch et al., 2015), semi-supervised learning (Chapelle et al., 2009), generative modeling (Don-ahue et al., 2017), and disentanglement learning (Higgins et al., 2017).”
S4L: Self-Supervised Semi-Supervised Learning (2019, background): “Given the wide variety of semi-supervised learning techniques proposed in the literature, we refer to [4] for an extensive survey.”
Bootstrap Your Own Latent: A New Approach to Self-Supervised Learning (2020, background): “In the semi-supervised setting [55, 56], an unsupervised loss is combined with a classification loss over a handful of labels to ground the training [19, 20, 57, 58, 59, 60, 61, 62].”
Your Classifier is Secretly an Energy Based Model and You Should Treat it Like One (2019, background): “1 INTRODUCTION For decades, research on generative models has been motivated by the promise that generative models can benefit downstream problems such as semi-supervised learning, imputation of missing data, and calibration of uncertainty (e.g., Chapelle et al. (2006); Dempster et al. (1977)).”
The Graph Neural Network Model (2009, neutral): “Those techniques include random fields [42], Bayesian networks [43], statistical relational learning [44], transductive learning [45], and semisupervised approaches for graph processing [46].”
Deep learning via semi-supervised embedding (2008, background): “…2005), label propagation (Zhu & Ghahramani, 2002) and LapSVM (Belkin et al., 2006), to name a few, make use of regularizers that are directly related to 1This is often referred to as the cluster assumption or the manifold assumption (Chapelle et al., 2006). unsupervised embedding algorithms.”
Regularization With Stochastic Transformations and Perturbations for Deep Semi-Supervised Learning (2016, background): “There are many approaches to semi-supervised learning in general [10, 11].”
Seeing through the Human Reporting Bias: Visual Classifiers from Noisy Human-Centric Labels (2015, background): “SSL approaches are often computationally impractical [56, 58] or make strong independence assumptions [13] that do not hold in human-centric annotations.”
Toward Causal Representation Learning (2021, neutral): “The view of SSL as exploiting dependencies between a marginal P ( X ) and a noncausal conditional P ( Y | X ) is consistent with the common assumptions employed to justify SSL [45].”
Learning privately from multiparty data (2016, methodology): “Furthermore, we are using unlabeled data to augment labeled data during training, which can be considered a semisupervised learning method (Chapelle et al., 2006).”
Causality for Machine Learning (2019, background): “I had previously worked on SSL (Chapelle et al., 2006), and it was clear that this was exactly the kind of information that SSL required when trying to improve the estimate of output give input by using unlabelled inputs.”
Semi-supervised learning, causality, and the conditional cluster assumption (2019, background): “estimate of P(X), but SSL aims at improving P(YjX). This can only work if there is a link between P(X) and P(YjX). Indeed, many approaches to SSL establish such a link through additional assumptions (Chapelle et al., 2010; Mey and Loog, 2019; Zhu, 2005). Two common ones are the cluster assumption, p”
Continual semi-supervised learning through contrastive interpolation consistency (2021, background): “sumptions on how input and labels interact [7].”
Learning from untrusted data (2016, methodology): “Our results have implications in a variety of domains, including building secure machine learning systems, performing robust statistics in the presence of outliers, and agnostically learning mixture models.”
Meta Pseudo Labels (2020, methodology): “We first study the role of feedback in Meta Pseudo Labels on the simple TwoMoon dataset [7].”
Interpolation Consistency Training for Semi-Supervised Learning (2019, methodology): “These methods are motivated by the low-density separation assumption (Chapelle et al., 2010), and as such push the decision boundary to lie in the low-density regions of the input space, achieving larger classification margins.”
Data Distillation: Towards Omni-Supervised Learning (2017, background): “There is a great volume of work on semi-supervised learning, and comprehensive surveys can be found in [49, 4, 50].”
A Comprehensive Survey on Transfer Learning (2019, background): “Definition 3 (Transfer Learning): Given some/an observation(s) corresponding to mS ∈ N+ source domain(s) and task(s) (i.e., {(DSi , TSi)|i = 1, . . . , mS}), and some/an observation(s) about mT ∈ N+ target domain(s) and task(s) (i.e., {(DTj , TTj )|j = 1, . . . , mT }), transfer learning utilizes th”