Essay details:

  • Subject area(s): Engineering
  • Price: Free download
  • Published on: 7th September 2019
  • File format: Text
  • Number of pages: 2

Text preview of this essay:

This page is a preview - download the full version of this essay above.

Using neural networks for translation

Online translators have improved drastically over the years and now, thanks to neural networks, they are improving at a faster rate.

Understanding foreign languages has always been a barrier for individuals and businesses seeking to expand in other countries. Language learning and translation has somewhat aided this and now with the recent technological advancements in neural networks, it may help us even more.

Erol Gelenbe is a professor at Imperial College London in the department of electrical and electronic engineering. His interest in neural networks came from trying to build mathematical models of parts of human and animal brains. He then started using neural networks to route data traffic in very large networks such as the internet.

Gelenbe says translation has three different aspects. The first is word to word translations, which can be accelerated or simplified using neural networks and other fast algorithms. The second is mapping the syntax, which means the neural network will have to “understand” grammar. The third is using context to translate, which is extremely important as it directly affects which words are chosen.

Gelenbe uses English and German as an example: “Neural networks can be used for each of these steps as a way to store and match patterns, for example matching ‘school’ with ‘schule’, matching ‘to’ with ‘nach’, or learning and matching the grammatical structures”.

Neural networks composed of many layers are referred to as “deep learning”. Gelenbe says: “It’s a special use of neural networks where many successive networks are being used on the same data, in alternating layers of networks. While one layer tries to search widely, the next layer specialises the findings through some optimisation scheme.

“In deep learning, these successive layers are repeated a large number of times. More conventional neural network learning will use only two layers, even though certain approaches use ‘recurrent networks’ which have feedback loops inside the networks to help learn complicated patterns.”

A robotic rosetta stone?

Google and Microsoft introduced neural machine translation back in November 2016. It’s different to the previous large-scale statistical machine translation, as it translates whole sentences at a time instead of each word, or a couple of words. In a blog post, Google explained how the sentence is translated in its broader context and is then rearranged and adjusted “to be more like a human speaking with proper grammar”. This makes it easier to translate larger bodies of text as they are taken sentence by sentence, so paragraphs and articles will be translated with fewer errors or instances of miscomprehension. Microsoft has a useful tool to highlight the difference between neural networks and statistical machine translation, where you can see how neural translation sounds much more natural. And the best part? Over time neural networks learn to create better and more natural translation.

Jürgen Schmidhuber is a professor and co-director of the Swiss Dalle Molle institute for artificial intelligence. He is also the president of NNAISENSE, a research team focused on building large-scale neural network solutions.

Since 2015, Google’s speech recognition on smartphones has been based on his self-learning  long short-term memory (LSTM) recurrent neural networks (RNNs). The technology was extended to Google Translate and Google\'s image caption generation as well as the new Google Allo assistant. Other companies such as Microsoft, IBM and Baidu also use his methods and even Apple is using LSTM on the iPhone. “Whenever you’re talking to Amazon Echo or Alexa, you are talking to an LSTM” he says.

Schmidhuber underlines that LSTM-based systems can carry out many other functions in addition to translations, including controlling robots, analysing images, summarising documents and predicting diseases.

What will the not so distant future of neural machine translation bring? “End-to-end video-based speech recognition and translation including lip-reading and face animation” Schmidhuber predicts.  

“For example, suppose you are in a video chat with your colleague in China. You speak English, he speaks Chinese. But to him it will seem as if you speak Chinese, because your intonation and the lip movements in the video will be automatically adjusted such that you not only sound like someone who speaks Chinese, but also look like it. And vice versa,” he explains.

Neural knowledge

Towards the end of April 2017, an update was released for Google Translate adding translation between English and nine Indian languages using neural machine translation technology. Previously, Translate supported translations between English, French, German, Spanish, Portuguese, Chinese, Japanese, Korean and Turkish using the new technology. This is a very important step for India as it is a country with 23 official languages.

<blockquote class=\"twitter-tweet\" data-lang=\"es\"><p lang=\"en\" dir=\"ltr\">A leap in translation accuracy: Google Translate, now with Neural Machine Translation for 9 Indian languages - <a href=\"\"></a> <a href=\"\"></a></p>&mdash; Google India (@GoogleIndia) <a href=\"\">25 de abril de 2017</a></blockquote>

<script async src=\"//\" charset=\"utf-8\"></script>

Google highlights that: “This new technique improves the quality of translation more in a single jump than we’ve seen in the last ten years combined”. Google has also discovered that neural technology speaks a language better when it learns several at a time, similar to how it\'s easier for humans to learn a language when they know a related one. This means if there isn’t a lot of sample data for one language, such as Bengali, but there is a lot for another, like Hindi, then the translation improves more when training them together.

Google Brain is a deep learning AI research project which investigates topics such as machine translation, natural language processing and quantum AI. Researchers at the project have recently announced they are using neural networks to aid language translation, but from speech to text. A user speaks the language which they want translated and it is then written in a different language straight away. This means that there is no transcription of the text in the source language.

By translating speech directly to text it can be utilised in a number of ways. This includes Robot Lawyer’s Do Not Pay, which was originally built to help people work out if they had to pay a parking fine or not and now also helps refugees apply for refugee status in foreign countries. Users could talk to the robot in their native language to facilitate their use of the tool. Another example is Babylon, an AI-powered health app that helps users diagnose themselves through a series of questions. By opening it up to other languages, it will be able to help a lot more people.

Improving translation

Rick Rashid founded Microsoft Research in 1991 and says that for nearly 10 years there was no change to word error rate in online translation, which is why deep learning and neural networks are so important.

Thanks to researchers at Microsoft, the word error rate in translations was improved drastically because of deep learning neural networks. As an example, Rashid says: “In 2012 I was able to stand up on stage in Tianjin, China and have my own voice simultaneous translated from English to Chinese live on stage. This was testament to the huge improvement in word error rates and to the translation technology we put in place.”

This technological jump could aid in the business world, where clients or colleagues don’t speak the same language. In particular, this can help SMBs grow faster and expand into areas of the world that were previously out of their reach.

It may have a huge impact on cross-industry learning too, as companies will be able to communicate better across borders. They can ask each other questions about projects they have carried out and then put it to use in their own business. More business models can be adapted and put to use in different industries to foster ground breaking change.

There is still much more to be explored with neural networks, Gelenbe still thinks the most exciting neural network discoveries await us

“To understand how our brain does very complicated things very quickly and so efficiently, is still before us -- the future will be more exciting than the past.”

...(download the rest of the essay above)

About this essay:

This essay was submitted to us by a student in order to help you with your studies.

If you use part of this page in your own work, you need to provide a citation, as follows:

Essay Sauce, . Available from:< > [Accessed 17.02.20].