Comparison and Cooperation of Several Classifiers

Driancourt, X.; Bottou, Léon; Gallinari, Patrick

doi:10.1016/b978-0-444-89178-5.50157-3

Cited by 2 publications

(3 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Viterbi training, though formulated differently, is often use in HMM-based speech recognition systems [28]. Similar algorithms have been applied to speech recognition systems that integrate NN's with time alignment [71], [72], [76] or hybrid neuralnetwork/HMM systems [29], [74], [75].…”

Section: A Viterbi Trainingmentioning

confidence: 99%

“…Variations of this technique have been used for the speech recognition. Driancourt and Bottou [76] used a version of it where the loss function is saturated to a fixed value. This can be seen as a generalization of the Learning Vector Quantization 2 (LVQ-2) loss function [80].…”

Section: B Discriminative Viterbi Trainingmentioning

confidence: 99%

“…The sum of those posteriors for all the possible label sequences is one. Another approach would consists of directly minimizing an approximation of the number of misclassifications [83], [76]. We prefer to use the discriminative forward loss function because it causes less numerical problems during the optimization.…”

Section: E Remarks On Discriminative Trainingmentioning

confidence: 99%

See 2 more Smart Citations

GradientBased Learning Applied to Document Recognition

Haykin

Kosko

2009

Intelligent Signal Processing

198

View full text Add to dashboard Cite

Multilayer neural networks trained with the back-propagation algorithm constitute the best example of a successful gradientbased learning technique. Given an appropriate network architecture, gradient-based learning algorithms can be used to synthesize a complex decision surface that can classify high-dimensional patterns, such as handwritten characters, with minimal preprocessing. This paper reviews various methods applied to handwritten character recognition and compares them on a standard handwritten digit recognition task. Convolutional neural networks, which are specifically designed to deal with the variability of two dimensional (2-D) shapes, are shown to outperform all other techniques. Real-life document recognition systems are composed of multiple modules including field extraction, segmentation, recognition, and language modeling. A new learning paradigm, called graph transformer networks (GTN's), allows such multimodule systems to be trained globally using gradient-based methods so as to minimize an overall performance measure. Two systems for online handwriting recognition are described. Experiments demonstrate the advantage of global training, and the flexibility of graph transformer networks. A graph transformer network for reading a bank check is also described. It uses convolutional neural network character recognizers combined with global training techniques to provide record accuracy on business and personal checks. It is deployed commercially and reads several million checks per day.

show abstract