This research work investigates design and analysis of an optimal classifier for the categorization of handwritten Marathi consonant characters of Devnagari script using a single hidden layer feed-forward neural network with five fold cross validation. Each neural network is trained three times by varying neurons in hidden layer from 64 to 128 in steps of 16. Scrupulous experimentation around seventy five MLPs shows the average classification accuracy is above 97% for all 32 classes. The best network with 128 neurons is further analyzed on account of confusion matrix, reveals the greater details for individual classes. Overall, classification accuracy on training, validation, test and combined dataset is 99.58%, 97.88%, 97.62% and 99.05% respectively on the total dataset size of 8224 samples distributed uniformly within 32 classes of typical Devnagari consonants.

No abstract

No abstract

No abstract

No abstract