Beyond the point cloud: from transductive to semi-supervised learning

By Vikas Sindhwani, Partha Niyogi, Mikhail Belkin. International Conference on Machine Learning, 2005

514 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/ec7c68427a26f812532b1c913c68fcf84b7de58e

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 Computer Vision and Pattern Recognition

Citations in the corpus

Semi-Supervised Learning via Generalized Maximum Entropy (2010, methodology): “Finally, even though our motivation is similar to other SSL methods that are based on the smoothness criterion, the resulting formulation is substantially different as it enables the algorithm to choose which similarities are salient unlike many SSL algorithms that treat similarities uniformly such ”
Semi-Supervised Multi-Task Regression (2009, background): “to the semi-supervised kernel function defined in [ 29 ].”
9 Entropy Regularization (background): “Hence, these results illustrate that manifold learning requires more tuning efforts for truly high dimensional data, and some recent techniques may respond to this need [Sindhwani et al., 2005].”
Semi-Supervised and Unsupervised Extreme Learning Machines (2014, methodology): “In the second stage, we followed the methods discussed in [52] and [57] to build the graph Laplacian L, and the hyperparameter settings can be found in [47], [52], and [57].”
Discriminative Semi-Supervised Feature Selection Via Manifold Regularization (2009, methodology): “Among these semi-supervised learning algorithms, the method of semi-supervised SVM with manifold regularization has demonstrated good performance [3], [45].”
Seeing stars when there aren’t many stars: Graph-based semi-supervised learning for sentiment categorization (2006, background): “We will extend it to the inductive learning setting based on (Sindhwani et al., 2005).”
Manifold Regularization: A Geometric Framework for Learning from Labeled and Unlabeled Examples (2006, methodology): “For several other experimental observations and comparisons on this data set, see Sindhwani et al. (2005).”
Learning Transferrable Representations for Unsupervised Domain Adaptation (2016, background): “In this regard, the research community has proposed different mechanisms such as semi-supervised learning [27, 37], transfer learning [23, 31], weakly labelled learning, and domain adaptation.”
Semi-supervised Discriminant Analysis (2007, background): “This formulation is motivated from spectral dimensionality reduction [2, 13], which also plays a key role in spectral clustering [17] and various kinds of graph based semisupervised learning algorithms [3, 6, 20].”
Semisupervised Neural Networks for Efficient Hyperspectral Image Classification (2010, background): “More details can be found in [13], [14], and its application to remote sensing data classification in [15].”
Introduction to Semi-Supervised Learning (2009, methodology): “THE ASSUMPTION OF GRAPH-BASED METHODS∗ 55 walk [8, 168], harmonic function [210], local and global consistency [198], manifold regularization [17, 158, 155], kernels from the graph Laplacian [35, 51, 95, 101, 163, 211], spectral graph transducer [90], local averaging [179, 187], density-based regula”
Mean Map Kernel Methods for Semisupervised Cloud Classification (2010, background): “In [41] and [42], labeled and unlabeled samples were”
Diving into the shallows: a computational perspective on large-scale shallow learning (2017, methodology): “The norm in kEP is a finite rank modification of the norm in the RKHS corresponding to k, a setting reminiscent of [SNB05] where unlabeled data was used to “warp” the norm for semi-supervised learning.”
Tighter PAC-Bayes bounds through distribution-dependent priors (2013, methodology): “We consider the “warped” RKHS (Sindhwani et al., 2005) H̃K of functions from HK with modified inner product, 〈h, g〉H̃K := 〈h, g〉K + τ〈h, g〉L, (41) where τ controls the relative weight given to the inner product inHK and the empirical inner product. The motivation here is that we are using the data t”
Value Regularization and Fenchel Duality (2007, background): “…(Rahimi et al., 2005) and structured prediction (Altun et al., 2005).8 In fact, whenever the additional regularization term is a quadratic form QL, we can view the semisupervised algorithm as finding the optimal function in a data-dependent “warped” RKHS (see Sindhwani et al. (2005) for details).”
Large Scale Transductive SVMs (2006, methodology): “Some of the main methods include (Chapelle et al., 2002; Chapelle and Zien, 2005; Sindhwani et al., 2005; Szummer and Jaakkola, 2001b); and (Weston et al., 2003).”