Support vector machine active learning for image retrieval

By Simon Tong, Edward Y. Chang. 2001

1,552 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

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

Learning Algorithms for Active Learning (2017, methodology): “For instance, Lewis & Gale (1994) and Tong & Chang (2001) developed policies based on the conﬁdence of the classiﬁer, while Gilad-Bachrach et al. (2005) used the disagreement of a committee of classiﬁers.”
Examining multiple potential models in end-user interactive concept learning (2010, background): “Prior work has thus focused interaction on prompting a person to answer “what class is this object?” [10, 7].”
Asymmetric bagging and random subspace for support vector machines-based relevance feedback in image retrieval (2006, background): “Among these classifiers, the Support Vector Machines (SVM) [8], [31], [27], [6], [25] based relevance feedback (SVM RF) has shown promising results owing to its good generalization ability.”
Image retrieval: Ideas, influences, and trends of the new age (2008, methodology): “Active learning using SVMs was proposed for relevance feedback [Tong and Chang 2001] and helped to popularize active learning in other domains as well.”
Applying Support Vector Machines to Imbalanced Datasets (2004, methodology): “Support Vector Machines (SVM) were introduced by Vapnik and colleagues [13] and they have been very successful in application areas ranging from image retrieval [12], handwriting recognition [3] to text classification [7].”
Active Learning Literature Survey (2009, background): “…and Koller, 2000; Hoi et al., 2006a), information extraction (Thompson et al., 1999; Settles and Craven, 2008), image classiﬁcation and retrieval (Tong and Chang, 2001; Zhang and Chen, 2002), video classiﬁcation and retrieval (Yan et al., 2003; Hauptmann et al., 2006), speech recognition (Tur…”
Multi-Class Active Learning by Uncertainty Sampling with Diversity Maximization (2015, methodology): “Compared with SVM active learning (Tong and Chang 2001), the minmax optimization algorithm is able to select training data in batch mode and is more robust to over-fitting.”
Active learning using pre-clustering (2004, methodology): “This simple and intuitive criterion performs well in some applications (Lewis & Gale, 1994; Tong & Chang, 2001; Schohn & Cohn, 2000; Campbell et al., 2000).”
Entropy-based active learning for object recognition (2008, positive): “This idea has been further developed for image retrieval experiments [10].”
Content-based image retrieval: approaches and trends of the new age (2005, positive): “Active learning using SVMs was introduced into the field of image retrieval in [95].”
KBA: kernel boundary alignment considering imbalanced data distribution (2005, methodology): “The image data set contains 20K images in 116 categories collected from the Corel Image CDs. 7 Each image is represented by a vector of 144 dimensions including color, texture, and shape features [22].”
Multi-class active learning for image classification (2009, background): “Tong and Chang [18] propose active learning for SVM in a relevance feedback framework for image retrieval.”
Class-Boundary Alignment for Imbalanced Dataset Learning (2003, methodology): “Notice that in this experiment, we use the L 1 -norm RBF function for D(x), since the L 1 -norm RBF works the best for the image dataset (Tong & Chang, 2001).”
A survey of content-based image retrieval with high-level semantics (2007, background): “and is considered a good candidate for learning in image retrieval system [35,69,81].”
Exploiting Unlabeled Data in Content-Based Image Retrieval (2004, methodology): “In the Svm Active approach developed by Tong and Chang [17], in each round of relevance feedback, a support vector machine is trained on labeled data and then the user is asked to label the images most close to the support vector boundary.”