Active + Semi-supervised Learning = Robust Multi-View Learning

By Ion Muslea, Steven N. Minton, Craig A. Knoblock. International Conference on Machine Learning, 2002

349 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/c84b9806d3de2d3b2d45f8fcb2df6760e98b1a15

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

Active Learning with Gaussian Processes for Object Categorization (2007, background): “The paradigm of active learning has been introduced in the machine learning community to address this issue [8, 26, 16, 18, 34]; with an active learning method, generally new test points are selected so as to minimize the model entropy.”
Semi-Supervised Active Learning for Sound Classification in Hybrid Learning Environments (2016, methodology): “The combined method proposed in [34] clearly outperformed co-EM, general co-testing and co-training in Web pages and pictures classification.”
Combining Committee-Based Semi-Supervised Learning and Active Learning (2010, methodology): “In [16], Co-Testing and Co-EM are combined, called CoEMT, to produce an active multi-view semi-supervised algorithm.”
Combining active learning and semi-supervised learning using Gaussian fields and harmonic functions (2003, background): “Exceptions include McCallum and Nigam (1998) who use EM with unlabeled data integrated into the active learning, and Muslea et al. (2002) who use a semi-supervised learning method during training.”
Introduction to Semi-Supervised Learning (2009, unknown): “, [85, 126, 132, 183, 203, 202, 213].”
A Survey on Multi-view Learning (2013, methodology): “Keywords: Multi-view Learning, Survey, Machine Learning”
Semi-supervised learning with graphs (2005, background): “Another exception is (Muslea et al., 2002), which uses a semi-supervised learning method during training.”
Multi-view clustering (2004, methodology): “If few labeled examples and, in addition, unlabeled data are available, then the co-training algorithm [4] and other multi-view classification algorithms [ 15 , 5] improve the classification accuracy often substantially.”
Multimodal Fusion for Video Search Reranking (2010, methodology): “Actually, this idea is derived from the multiview learning strategy [23], [24], [25], [26], a semisupervised method in machine learning.”
Building text classifiers using positive and unlabeled examples (2003, background): “Finally, our work is related to learning using a small labeled set and a large unlabeled set [2][3][4][5][10] [11][25][26][28][35].”
Gaussian Processes for Object Categorization (2010, background): “…learning has been introduced in the machine learning community to address this issue (Freund et al. 1997; Tong and Koller 2000; McCallum and Nigam 1998; Muslea et al. 2002; Zhu et al. 2003); with an active learning method, generally new test points are selected so as to minimize the model entropy.”
Discriminative Batch Mode Active Learning (2007, background): “Furthermore, if a parallel labeling system is available, a single instance selection system can make wasteful use of the resource.”
Bootstrapping Coreference Classifiers with Multiple Machine Learning Algorithms (2003, methodology): “In addition, we investigate a method for ranking unlabeled instances to be fed back into the bootstrapping loop as labeled data, aiming to alleviate the problem of performance deterioration that is commonly observed in the course of bootstrapping.”
Semi-Supervised Active Learning for Sequence Labeling (2009, methodology): “Similarly, co-testing (Muslea et al., 2002), a multi-view AL algorithms, selects examples for the multi-view, semi-supervised Co-EM algorithm.”
Learning With Feature Evolvable Streams (2017, background): “ procedure. However, most of existing online learning algorithms are limited to the case that the feature set is ﬁxed. Other related topics involving multiple feature sets include multi-view learning [19, 20, 33], transfer learning [24, 25] and incremental attribute learning [11]. Although both our ”