Machine Translation

Introduction Although the field of computational linguistics is relatively new, it contains several subfields that reflect the practical application of that field. Machine translation (MT) is one of the main components of Computational Linguistics (CL). People working in this field are not necessarily experts in other fields of CL, so they can be regarded as independent tasks. But connecting them is that all these topics use computers as tools for dealing with human languages.

Machine translation (MT) automatically translates text from one natural language (source language) to another language (target language) using machine capabilities. The idea of ​​using a machine for translation was first proposed by Warren Weaver in 1949. Machine translation has been done for a long time (from the 1950's to the 1980's) by examining the language information of the source language and the translation language, and generating a translation based on the dictionary. And the grammar is called Rule Based Machine Translation (RBMT). Along with the development of statistics, statistical models became applied to machine translation, and translation was generated based on an analysis of bilingual text corpus. This approach, known as Statistical Machine Translation (SMT), achieved superior performance than RBMT and dominated over the 1980s and early 2000s. Their research laid the foundation for the future application of neural networks in machine translation.

Machine translation usually uses phrase-based statistical methods, but neural networks are rapidly becoming dominant. My favorite architecture is trained in a multilingual environment where a single model has been translated from any source language to any target language and has a weakly overlooked (or not fully supervised) setting. Game Although the clearly written Go program has been developed for a long time, AlphaGo Zero, ConvNet, which examines the original state of the board, is currently the most powerful player in the game. I hope that you will get very similar results in other fields such as DOTA 2 and StarCraft.