Improving the Accuracy and Speed of Support Vector Machines

By C. Burges, Bernhard Schölkopf. 1996

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

Dropout Training as Adaptive Regularization (2013, background): “Dropout training falls into the broader category of learning methods that artificially corrupt training data to stabilize predictions [2, 4, 5, 6, 7].”
The kernel recursive least-squares algorithm (2004, background): “Burges [13], [14] seems to have been the first”
Sparse Bayesian Learning and the Relevance Vector Machine (2001, background): “Some form of post-processing is often required to reduce computational complexity (Burges, 1996; Burges and Scholkopf, 1997).”
Shape Matching and Object Recognition Using Shape Contexts (2002, methodology): “Some examples are the LeNet classi er [34], a convolutional neural network for handwritten digit recognition, and the Support Vector Machine (SVM)-based methods of [41] (for discriminating between templates of pedestrians based on 2D wavelet coeÆcients) and [11, 7] (for handwritten digit recognition”
Marginalized Denoising Auto-encoders for Nonlinear Representations (2014, methodology): “This prior knowledge has been exploited to generate additional training samples for SVM classifiers or neural networks to improve generalization to unseen samples (Bishop, 1995; Burges & Schölkopf, 1997; Herbrich & Graepel, 2004; Ciresan et al., 2012).”
Bayesian Optimization with Inequality Constraints (2014, methodology): “…evaluated by uniformly sampling. a particular parameter setting γ 2 , C, m , an evaluation of () and c() involves first compressing an SVM with param- eters γ 2 , C down to m support vectors following Burges & Schölkopf (1997), and then evaluating the resulting classifier on the validation set.”
Object Recognition with Gradient-Based Learning (1999, background): “The Support Vector Machine [31] has excellent accuracy, which is most remarkable, because unlike the other high performance classiers, it does not include ap rioriknowledge about the problem [ 4 ].”
Random features for Kernel Deep Convex Network (2013, methodology): “(i) Burge and Schölkopf proposed a reduced set method by finding a smaller number training set in order to improve classification speed [7].”
Efficient shape matching using shape contexts (2005, unknown): “These approaches have been used for handwritten digit recognition [10], [11], face recognition [12], and isolated 3D object recognition [13].”
Churn prediction in subscription services: An application of support vector machines while comparing two parameter-selection techniques (2008, background): “, 2005) and hand-written digit recognition (Burges and Scholkopf, 1997; Cortes and Vapnik, 1995), the applications in marketing are rather scarce (Cui and Curry, 2005).”
Learning Nonlinear Functions Using Regularized Greedy Forest (2011, background): “ave thoroughly explored the potential of AdaBoost in our experiments. RGF’s 1.4% exceeds a number of neural networks in [LeCun et al., 1998], and it is as good as SVM with 4-degree polynomial kernel [Burges and Scholkopf, 1997], though it falls behind the convolutional nets LeNet-4 and¨ -5 (1.1%) [L”
Image Retrieval Using Textual Cues (2013, background): “Note that the use of affine transformations in training examples is shown to improve classification accuracy [14, 20].”
Data Augmentation via Levy Processes (2016, unknown): “One popular strategy for encoding such a belief is data augmentation: generating additional pseudo-examples or “hints” by applying label-invariant transformations to training examples’ features (Abu-Mostafa, 1990; Schölkopf et al., 1997; Simard et al., 1998).”
Gradient-based learning applied to document recognition (1998, methodology): “1% on the regular test set [63], with a computational cost of only 650 000 multiply–adds per recognition, i.”