Deep Learning Model For Sequential Data -Machine Language Translation

Abstract

This paper discusses Deep Neural Networks (DNN) and deep learning as it relates to machine translation, form of natural language processing. DNN is now a key component of machine learning methodologies. One of the best techniques for machine learning is the recursive recurrent neural network (R2NN). Recursive and recurrent neural networks are combined to create it (such as Recursive auto encoder). In this research, semi-supervised learning techniques are used to train the LSTM for reordering from source to target language. To create word vectors for the source language, the Seq2word tool is necessary, and the auto encoder aids in the reconstruction of the vectors for the destination language in a tree structure. The output of seq2word is crucial for the input vectors'
word alignment. Due to the LSTM structure's complexity and time required to train the enormous data set using seq2seq. The performance accuracy is anlayzed using BLEU score