Fast support vector machine training and classification on graphics processors

By Bryan Catanzaro, Narayanan Sundaram, Kurt Keutzer. International Conference on Machine Learning, 2008

415 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/2635f61333900a6b4cd9b5db5d4c3bc31363b2ff

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

Empowering Visual Categorization With the GPU (2011, methodology): “However, such a naive implementation is not very efficient, because the same result is obtained with fewer operations by simply vectorizing the Euclidean distance, which is a common trick [26]:”
Compute Trends Across Three Eras of Machine Learning (2022, methodology): “CNNs were trained on GPUs at least as early as 2006 (Chellapilla et al., 2006), as were some other models that were large at the time (such as the 100M parameter Deep Belief Network by Raina et al. (2009a) and the neural networks of D. C. Cire¸san et al. (2011)).”
SystemML: Declarative machine learning on MapReduce (2011, background): “Previous attempts at building scalable machine learning algorithms have largely been hand-tuned implementations on specialized hardware/parallel architectures [1], or as noted in [2], clever methods to parallelize individual learning algorithms on a cluster of machines [3], [4], [5].”
Materials that couple sensing, actuation, computation, and communication (2015, methodology): “In [73], a fully distributed algorithm for training an SVM is described which leverages a geometric interpretation of SVMs to minimize memory and communication requirements, and trains a distributed SVM using a gossip protocol.”
A Practical and Conceptual Framework for Learning in Control (2010, methodology): “Catanzaro et al. (2008) used them in the context of support vector machines whereas Raina et al. (2009) applied them to large-scale deep learning.”
Large-scale deep unsupervised learning using graphics processors (2009, methodology): “GPUs have been applied to certain problems in machine learning, including SVMs (Catanzaro et al., 2008), and supervised learning in convolutional networks (Chellapilla et al.”
Kernel Methods for Deep Learning (2009, methodology): “, 2005), parallelization of support vector machines (Catanzaro et al., 2008), and approximation of kernel operations with explicit nonlinear mapping (Rahimi and Recht, 2008) are representative work in this direction, and we are considering these methods to make arc-cosine kernels more scalable to la”
Predicting the Price of Bitcoin Using Machine Learning (2018, background): “Similarly, [24] reported an increased speed in classification time to the magnitude of eighty times when implementing a SVM on a GPU over an alternative SVM algorithm run on a CPU.”
Adaptive Kernel Approximation for Large-Scale Non-Linear SVM Prediction (2011, methodology): “The SVM prediction process may become computationally demanding (Joachims & Yu, 2009; Rai et al., 2009; Catanzaro et al., 2008) especially for large-scale data collections with complicated decision boundaries, such as image and audio data, where the size of SVs is not much smaller than the training…”
Tensor-Based Learning for Predicting Stock Movements (2015, neutral): “The paralleled SVR processing is a promising way to deal with the scalability problem (Catanzaro, Sundaram, and Keutzer 2008).”
A Tensor-Based Information Framework for Predicting the Stock Market (2016, background): “Parallel SVR processing is a promising approach to addressing the scalability problem [Catanzaro et al. 2008]; however, its effectiveness for the proposed tensor-based information framework has yet to be explored.”
Spectral–Spatial Classification of Hyperspectral Imagery Based on Partitional Clustering Techniques (2009, neutral): “However, in recent works, methods and parallel implementations to speed up the SVM training and classification have been proposed [64], [69], [70], [ 71 ].”