Skip to main content current edition: International edition The Guardian - Back to home Become a supporter Subscribe Find a job Jobs Sign in Search Show More Close with google sign in become a supporter subscribe search find a job dating more from the guardian: change edition: edition International edition The Guardian - Back to home browse all sections close Google Google buys neural network city guide creator Jetpac Artificial intelligence-based method of creating guides to cities using public photos appealed to Google’s drive for building smart systems Belfast at dusk - identified as Britain's happiest city according to Jetpac's neural network analysis of public photos.
 to Jetpac’s neural network analysis of public photos.
 Photograph: scenicireland.
com/Christopher/Alamy Google Google buys neural network city guide creator Jetpac Artificial intelligence-based method of creating guides to cities using public photos appealed to Google’s drive for building smart systems Charles Arthur @charlesarthur Mon 18 Aug ‘14 14.
35 BST Last modified on Tue 21 Feb ‘17 18.
33 GMT This article is 3 years old Google is buying Jetpac, a “city guides” company with a twist which used image recognition and neural network technology to recommend places it deemed the happiest, most popular or with the best views and scenic hikes.
 Jetpac offered special “City Guides” for more than 6,000 destinations, using neural network technology developed by Pete Warden, the company’s co-founder and chief technology officer.
 “We can spot lipstick, blue sky views, hipster moustaches and more, through advanced image processing on billions of photos,” Jetpac’s home page explains.
 The app worked by analysing public photos with location data shared on Flickr, Instagram and other photo networks for particular elements, and then extracting key elements about them.
 The current Jetpac apps will be removed from the Apple App store within days, and support will end on 15 September, Jetpac says on its web page.
 It did not yet have an app for Android.
 The purchase, for an undisclosed sum, points to Google’s growing interest in artificial intelligence applications as it seeks to grow offerings such as its Google Now personal assistant.
 This year it acquired the British AI company DeepMind for $400m.
 Jetpac, with its neural network systems, seems to fit into that area.
 Warden wrote two demonstrator apps, currently still available on Apple’s App Store: “Spotter” which attempts to identify objects, and Deep Belief, which can be “trained” to recognise objects.
 Both use neural network systems for their processing.
 Neural networks are collections of algorithms which in effect mimic the functioning of brain systems: they can be “trained” to recognise particular elements in pictures, or in text, and flag their occurrence or absence.
 The system relies for its accuracy on feedback rather than explicit programming - so that in training a neural network to recognise a moustache, one would give it a huge number of photos to work on, and keep telling it when it was correct and incorrect.
 The neural network dynamically adjusts the weighting of various algorithms until it the answers are more and more correct.
 Earlier this year Jetpac said it had identified the “UK’s happiest city” - which it identified as Belfast, based on public Instagram photos.
 At the time Warden said that 5m photos per day were being uploaded which included geotagging and were public - the raw “data exhaust” that Jetpac needed to produce its analysis.
 In August 2012 Google bought the printed travel guide company Frommer’s, but then sold it back to the founder Arthur Frommer in April strategy Topics Loading comments… Trouble loading?
 most viewed The Guardian back to top all sections close back to top All rights reserved.
 ting%2CApps]