Performance Comparison of Statistical vs. Neural-Based Translation System on Low-Resource Languages

Abstract One of the important applications for which natural language processing (NLP) is used machine translation (MT) system, automatically converts one to another. It has witnessed various paradigm shifts since its inception. Statistical (SMT) dominated MT research decades. In recent past, researchers have focused on developing systems based artificial neural networks (ANN). this paper, first, some deep learning models that are mostly exploited in Neural Machine Translation (NMT) design discussed. A systematic comparison was done between performances SMT and NMT concerning English-to-Bangla English-to-Hindi tasks. Most Indian scripts morphologically rich, availability a sufficient corpus rare. We presented analyzed our work survey conducted other low-resource languages, finally useful conclusions been drawn.