Perceptron variants on OCR and Netflix datasets
نویسنده
چکیده
Since the perceptron was invented in 1957 by Rosenblatt, a large number of variants of the perceptron algorithm have been proposed. For this project, the main task was to find perceptron variants that worked well on the two different types of data sets, OCR and Netflix. Two variants, voted perceptron and longest survivor perceptron, were examined. Both variants were tested with several combinations of the methods, including the perceptron with margin and three different kernels, polynomial, Gaussian and the conventional dot product kernel. The results are compared and analyzed in detail.
منابع مشابه
Evaluation of pattern classifiers for fingerprint and OCR applications
In this paper we evaluate the classiication accuracy of four statistical and three neural network classiiers for two image based pattern classiication problems. These are ngerprint classiication and optical character recognition (OCR) for isolated handprinted digits. The evaluation results reported here should be useful for designers of practical systems for these two important commercial appli...
متن کاملEvaluation of Pattern Classiiers for Fingerprint and Ocr Applications
In this paper we evaluate the classiication accuracy of four statistical and three neural network classiiers for two image based pattern classiication problems. These are ngerprint classiication and optical character recognition (OCR) for isolated handprinted digits. The evaluation results reported here should be useful for designers of practical systems for these two important commercial appli...
متن کاملUsing HMMs to boost accuracy in optical character recognition
One of the current trends in the eld of optical character recognition (OCR) is the combination of multiple classi ers to produce more robust recognition. In this paper, a kNN classi er, a multi-layer perceptron, and a support vector machine with a radial basis function kernel are tested on a pre-segmented word-based OCR task with and without Viterbi errorcorrection. I discuss the effects of err...
متن کاملAn Incremental Framework Based on Cross-Validation for Estimating the Architecture of a Multilayer Perceptron
We define the problem of optimizing the architecture of a multilayer perceptron (MLP) as a state space search and propose the MOST (Multiple Operators using Statistical Tests) framework that incrementally modifies the structure and checks for improvement using cross-validation. We consider five variants that implement forward/backward search, using single/multiple operators and searching depth-...
متن کاملComparison of Performance of Variants of Single layer Perceptron Algorithms on Non separable Datasets
We present a detailed experimental comparison of the pocket algorithm thermal percep tron and barycentric correction procedure algorithms that most commonly used algorithms for training threshold logic units TLUs Each of these algorithms represent stable variants of the standard perceptron learning rule in that they guarantee convergence to zero classi cation errors on datasets that are linearl...
متن کامل