Hierarchical Convolutional Neural Network for Handwritten Digits Recognition
Zufar Kayumov, Dmitrii Tumakov, Sergey Mosin
Abstract
The application of a combination of convolutional neural networks for the recognition of handwritten digits is considered. Recognition is carried out by two sets of the networks following each other. The first neural network selects two digits with maximum activation functions. Depending on the winners, the next network is activated, which selects one digit from two. The proposed algorithm is tested on the data from MNIST. The minimal handwriting recognition error was estimated with this approach.
Topics & Concepts
MNIST databaseComputer scienceConvolutional neural networkDigit recognitionHandwriting recognitionPattern recognition (psychology)HandwritingSpeech recognitionArtificial neural networkNeocognitronArtificial intelligenceNumerical digitTime delay neural networkFeature extractionArithmeticMathematicsNeural Networks and ApplicationsHandwritten Text Recognition TechniquesImage Processing and 3D Reconstruction