Kannada to English Machine Translation Using Deep Neural Network

This paper proposes an NMT model which takes Kannada text as input and translates to the corresponding English using a sequence-to-sequence architecture with LSTM layers. The mechanism of encoder-decoder facilitates capturing and holding onto sentence context and, therefore, improves the accuracy of translation. Accuracy of 90.32\% was obtained by our approach. In the future work, Gated Recurrent Units and other transformer architectures may be extensively investigated for further improvement in Kannada and other low-resource language translation.