Towards optimal shallow ANN for recognizing isolated handwritten Bengali numerals

This work attempts to find the most optimal setting for shallow artificial neural network (ANN) for Bengali digit dataset. Recognition of handwritten Bengali numerals has recently gained much interest among researchers due to significant performance gain found in the recognition of English numerals using artificial neural network. In this work, a new dataset of 70,000 samples were created first by taking handwriting of 1750 persons where 982 persons were male and the rests were female. These individual image samples are then converted to grayscale, normalized, inverted and pickled to complete the data preprocessing step. Later this dataset was recognized using several artificial neural network settings where the most optimal setting being found to be one hidden layer with rectified linear unit activation and one output layer with softmax activation. In this study, we have given enough evidence to support our choice of these activations. We proposed the optimal number of neurons that can be used in the hidden layer is 700; also the optimal batch size is 200. The maximum accuracy on the test set has been found to be 96.05% which is a good result due to the complexity of the recognition task.