Artificial intelligence is getting shrewd in real time with massive amounts of computational power and high in demand in last quarter more than billion dollars. The Intensification of Neural Networks in Machine learning playing major revolution in Machine Translation (MT). Looking same aspect, we have prepared a paper on Machine Translation for Hindi using Neural Machine Translation techniques. The Basis of this paper is to translate the source language into target language with the help of sentence structure for source language (English sentence) and corresponding reordering rules for target language (Hindi sentence) using a deep neural network (DNN).When the source input that is English sentences are exactly matched with the target language database then the suitable translation for related input is directly fetched from the database. The fuzzy matching between the input sentence and the sample sentence in the database is done with cosine similarity between the sentence structure of input sentence and the sample sentence and then the reordering rule of the most alike sentence to the input sentence is used for translation.Extracting the structure of the source language sentence and extracting the reordering rule for target sentence is the core part of the translator. Reordering of words in spite of that remains one of the complex problems.It is exciting to implement Deep Neural Network and fuzzy logic to build intelligent reordering rules machine learning system.

A Part of Speech classifier is an important tool that is used to develop many NLP tasks. In this paper we described Deep Neural Network based architecture, that given a sentence, outputs a part-of-speech tag sequence. For Part of Speech classification, we trained a LSTM network by semi-supervised learning. We further applied a probabilistic approach to improve Part of Speech prediction of the LSTM network. Word2vec deep learning tool is used for learning high-quality distributed vector representations of words that capture a large number of precise word relationships. By using a small amount of labeled dataset of words and applying cosine similarity we assign Part of Speech tags to words. If some words don't show nearness to any word in labeled dataset, then we apply probabilistic approach for classification of unclassified words of input sentence. This hybrid neural Part of Speech classifier is evaluated over a corpus of 1000 sentences having a total of 7,400 words; part of speech classification system achieved the accuracy of 94.71 %.