This paper presents a recurrent neural networks applied to handwriting character recognition. The method Multi-dimensional Recurrent Neural Network is evaluated against classical techniques. To improve the model performance we propose the use of specialized Support Vector Machine combined whit the original Multi-dimensional Recurrent Neural Network in cases of confusion letters. The experiments were performed in the C-Cube database and compared with different classifiers. The hierarchical combination presented promising results. © 2012 Springer-Verlag.
CITATION STYLE
Bezerra, B. L. D., Zanchettin, C., & De Andrade, V. B. (2012). A MDRNN-SVM hybrid model for cursive offline handwriting recognition. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7553 LNCS, pp. 246–254). https://doi.org/10.1007/978-3-642-33266-1_31
Mendeley helps you to discover research relevant for your work.