In this section, we will talk about current trends in NLP. These trends are from the NLP research conducted between 2012 and early 2018. First let's talk about the current states of word embeddings. Word embeddings is a crucial topic as we have already seen many interesting tasks that rely on word embeddings to perform well. We will then look at important improvements in NMT.
Many variants of word embeddings have emerged over time. With the inception of high-quality word embeddings (refer to Distributed representations of words and phrases and their compositionality, Mikolov and others [1]) in NLP, it can be said that NLP had a resurgence, where many took an interest in using word embeddings in various NLP tasks (for example, sentiment analysis, machine translation, and question answering). Also, there have been many attempts to improve word embeddings, leading to even better embeddings. The four models that we'll introduce are in the areas of region embedding...