Word prediction using CNN for ancient manuscripts.

In this paper, we propose a CNN model for character recognition in Tamil palm-leaf manuscripts. Palm leaves served as the primary source for retaining information in numerous domains such as science, literature, medicine, history and so on in ancient times. Tamil is one of the ancient languages, and there are several manuscripts that provide important information. These manuscripts were destroyed of some bacteria. As a result, by predicting the missing letter and providing suggestions for identifying the word with a high prediction rate, this Convolutional neural network can play a significant role in training the data set and recovering valuable information from the manuscripts. [ABSTRACT FROM AUTHOR]