Neural Market Trends |||

Unsupervised Machine Learning Translation

I always wondered how Facebook does its in post’ language translation, and now I know.


German TranslationGerman Translation

Our new approach provides a dramatic improvement over previous state-of-the-art unsupervised approaches and is equivalent to supervised approaches trained with nearly 100,000 reference translations. To give some idea of the level of advancement, an improvement of 1 BLEU point (a common metric for judging the accuracy of MT) is considered a remarkable achievement in this field; our methods showed an improvement of more than 10 BLEU points.

It would appear that this is a multi-step process starting with a word-by-word translation, language modeling, then back translation.

In our research, we identified three steps — word-by-word initialization, language modeling, and back translation — as important principles for unsupervised MT.

Then they put it into an unsupervised neural model’ that did ok and was used for the back translation sentences.

The first one was an unsupervised neural model that was more fluent than word-by-word translations but did not produce translations of the quality we wanted. They were, however, good enough to be used as back-translation sentences. With back translation, this method performed about as well as a supervised model with 100,000 parallel sentences. via F code

All this reminds me of when Google’s machine learning process created it’s own’ interface language for translating languages.

Overall, pretty darn cool if you ask me.

Up next Understanding Random Forests with Python Code Towards Data Science is such a great Medium blog that I should consider writing for it! I came across a great article about Python Script to Parse Blog Feeds Version 2 I’m going to share an update to my original python script. It’s super simple and it completely automates parsing multiple blog feeds for you to
Latest posts Democratising Machine learning with H2O — Towards Data Science Getting started with Python datatable | Kaggle Phone Addiction Version 12 Launches Today! Machine Learning Making Pesto Tastier 5 Dangerous Things You Should Let Your Kids Do The Pyschology of Writing TensorFlow and High Level APIs Driving Marketing Performance with H2O Driverless AI Machine Learning and Data Munging in H2O Driverless AI with datatable Making AI Happen Without Getting Fired Latest Musings from a Traveling Sales Engineer The Night before H2O World 2019 Why Forex Trading is Frustrating Functional Programming in Python Automatic Feature Engineering with Driverless AI Ray Dalio's Pure Alpha Fund What's new in Driverless AI? Latest Writings Elsewhere - December 2018 House Buying Guide for Millennials Changing Pinboard Tags with Python Automate Feed Extraction and Posting it to Twitter Flux: A Machine Learning Framework for Julia Getting Started in Data Science Part 2 Makers vs Takers How Passive Investing Saved My Life Startups and Open Source The Process of Writing H2O AI World 2018 in London Ray Dalio's Pure Alpha Fund Isolation Forests in