Google has come up with a new computer program that is capable of navigating the London Underground system all by its own.
The first thing that pops into everyone’s head when they first hear about this is “My phone can do this with the help of Maps.” Well, this new program does the same thing with a different approach called Deep Learning. The system combines the data processing with self-learning code which will deliver the reasoning power similar to that of a human.
The new program does not need any aid from a human during the navigation and this is achieved by combining an external memory with deep-learning. The purpose of the external memory is to store pieces of information which are important and to access it whenever required. This is in fact compared to that of a normal human process where we access bits of information whenever we try to follow a process that involves different steps, such as a recipe.
The system has been successful in finding the shortest paths between the underground stops under tests. It was also able to answer where it would end up after we travel for two stops, like where we will be two stops from Victoria station.
The new program was given other tasks too, which included providing it with short statements such as “John is in the playground. John picked up the football.” and it responded accurately to the follow-up question asked to it like where the football was now.
Higher limits are now being reached with the help of Google’s Deepmind, which was able to beat the top player, Lee Sedol at the board game ‘Go’. Deep Learning is also able to perform complex processes such as image and speech recognition, language translation among many other computer strategies in a task.