Building a Robust Machine Translation System for Punjabi to English Using Deep Learning Approaches

In the dynamic realm of artificial intelligence, the evolution of Machine Translation Systems (MTS) for Indian languages faces unique challenges due to linguistic diversity. The diversity of morphology and the inherited dialects make it difficult to translate. The translation can be improved with pre-processing and optimization of MTS. Our MTS proposes the steps to improve the translation of Punjabi to English by employing two models, one with a Gated Recurrent Unit (GRU) layer and another with a Long Short-Term Memory (LSTM) layer. Another step involves Hyperparameter tuning to reduce complexity and increase model's efficiency. The experimentation shows accuracy rate on Samanantar dataset. The research aims to provide solutions to gaps identified in MTS for Punjabi, contributing to the advancement of machine translation technology in Punjabi's linguistic landscape.