SwiftKey improves its keyboard predictions with neural networks

SwiftKey has wrapped up its almost year-long pilot of a “neural network” keyboard and is bringing it to the Android masses.

Starting today, keyboard app SwiftKey will be running on a neural network. For example, having encountered the phrase “Let’s meet at the airport” during training, the technology can infer that “office” or “hotel” are similar words that could be appropriate replacements for “airport”.

The update is available today on Google Play.

The neural network technology was first used in an early experimental prototype, announced last October.

The keyboard also has the ability to understand how two sentences can have a similar structure. If you want to give this a try and see if it can save you a few keystrokes, then grab the latest version of SwiftKey from the Play Store.

SwiftKey Keyboard replaces a mobile device’s keyboard with one that understands language and learns your personal writing style.

To achieve this, SwiftKey has trained its app with millions of sentences.

Android owners can now download the app, initially available in US English and UK English, with more languages to come.

The company claims that this technique will not only improve word suggestions but also the “autocorrect” feature offered by the app.

However, Google has also used a mini neural network its Google Translate app, which enabled text to be translated on the screen via the camera lens even without an internet connection. SwiftKey’s original technology revolutionized touchscreen typing with the introduction of a row of next-word predictions – a feature that has since become the industry standard.

The London-based company gave its first public demonstration of the technology past year through a new standalone app called the SwiftKey Neural Alpha app, which was effectively an experimental version of its main app powered by a different language engine.