Cluster Kernels for Semi-Supervised Learning

By Olivier Chapelle, Jason Aaron Edward Weston, Bernhard Schölkopf. Neural Information Processing Systems, 2002

559 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/6bbc0c752570c46a772f2982728f9ad4191f25dd

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

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].”
Iterative Quantization: A Procrustean Approach to Learning Binary Codes for Large-Scale Image Retrieval (2013, unknown): “it is possible to scale the eigenvectors by the tth power of the eigenvalues [6], but in our experiments, we have found that t 1⁄4 1 works the best.”
Nonlocal Estimation of Manifold Structure (2006, background): “In recent years, there has been a lot of work on unsupervised learning based on characterizing a possibly nonlinear manifold near which the data would lie, such as locally linear embedding (LLE) (Roweis & Saul, 2000), Isomap (Tenenbaum, de Silva, & Langford, 2000), kernel principal component analysi”
Semi-Supervised Learning (Adaptive Computation and Machine Learning) (2006, background): “Indeed, Chapelle et al. [2002] and Smola and Kondor [2003] both suggest a general principle for creating a semi-supervised kernelK from the graph Laplacian.”
Transductive Learning via Spectral Graph Partitioning (2003, methodology): “We use D ii = i 2 (see also (Chapelle et al., 2002)).”
Beyond the point cloud: from transductive to semi-supervised learning (2005, methodology): “In particular, we note ( Chapelle, Weston & Schoelkopf, 2003; Chapelle & Zien, 2005) where kernels are designed to implement the cluster assumption.”
Hyperparameter Learning for Graph Based Semi-supervised Learning Algorithms (2006, methodology): “In the second step, smoothing is applied globally, typically, based on the spectral transformation of the graph Laplacian [6] [10].”
Combining Label Propagation and Simple Models Out-performs Graph Neural Networks (2020, background): “In the semi-supervised learning literature, the analog is the smoothness or cluster assumption (Chapelle et al., 2003; Zhu, 2005).”
Spectral Algorithms for Supervised Learning (2008, background): “Regularization has a long history in learning, and our starting point is the theoretical analysis proposed in Bauer, Pereverzev, and Rosasco (2007), De Vito, Rosasco, and Verri (2005), Caponnetto and De Vito (2007) , and Caponnetto (2006), showing that many regularization methods originally propose”
Transfer String Kernel for Cross-Context DNA-Protein Binding Prediction (2016, background): “(4) Under the semi-supervised setting, the so-called “sequence neighborhood” kernel or “cluster” kernel [18] proposes a semi-supervised extension of string kernel, which replaces every sequence with a set of “similar” (neighboring) sequences and a new representation is obtained by averaging represen”
A Comprehensive Survey on Test-Time Adaptation Under Distribution Shifts (2023, unknown): “explicitly or implicitly made in virtually all semi-supervised learning algorithms is the cluster assumption [146], [277], [278], which states that the decision boundary should not cross high-density regions, but instead lie in low-density regions.”
Semi-Supervised Graph-Based Hyperspectral Image Classification (2007, methodology): “supervised learning and clustering [35], [37], [46], and more specifically by the work of Zhu et al.”
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”
Analyzing Co-training Style Algorithms (2007, background): “…to be obtained in many practical machine learning applications, so semi-supervised learning which attempts to exploit unlabeled data to help improve the performance of learning with limited labeled training data has attracted much attention during the past few years [5, 9, 13, 6, 12, 10, 17, 3].”
Semi-Supervised Learning on Riemannian Manifolds (2004, methodology): “Now one might ask, how far is Êλ,n from Êλ? In order to get a handle on this question, one may proceed by building on the techniques described in Cucker and Smale (2001), Bousquet and Elisseeff (2001), and Kutin and Niyogi (2002).”
Semi-Supervised Classification by Low Density Separation (2005, methodology): “It has been implemented and used in Isomap [19], cluster kernels [7], and connectivity clustering [10].”
Consistency of spectral clustering (2008, background): “in Chapter 5 of Chatelin (1983) instead of the results of Atkinson (1967) (note that in the unnormalized case, the assumptions of Theorem 7 are not satisfied as we only have compact convergence instead of collectively compact convergence).”
Semi-Supervised and Unsupervised Deep Visual Learning: A Survey (2022, background): “The main assumptions for SSL include the smoothness assumption [41] and manifold assumption [8], [41] – the latter is also known as cluster assumption [42], structure assumption [43], and low-density separation assumption [44].”