One of the things we can do with technologies nowadays is to build a networks of simulated neurons, taking inspiration from biology. Facebook’s Chief Technology Officer confirms the fact that AI Research team is conducting research in areas like image recognition and natural language understanding.
FB’s team predict that one day AI systems are going to act as extensions of our brains, amplifying everything we do, augmenting our memories and giving us instant knowledge.
One of team’s ambition is to train machines to detect objects from photos or videos which initially they comprehend as pixels. This is called ‘segmentation’ – or the ability of the systems to segment images in order to distinguish the different objects on them. Their research agenda is going to be presented at next NIPS (Neural Information Processing Systems) meeting this year.
Neural information processing is a field which benefits from a combined view of biological, physical, mathematical, and computational sciences.
New technology called memory networks or MemNets are type of short-term memory which empowers deep-learning systems to achieve full natural language understanding. In simple words the aim of this research is to enable people to interact with their devices on a whole new level where it can interpret the context of your question better.
Long term goals include unsupervised and predictive learning, where the systems can learn through observation (instead of through direct instruction, which is known as supervised learning) and then begin to make predictions based on those observations.
One of the most common approaches in AI research to train machines is using games. For example the best human Go players often take advantage of their ability to recognise patterns on the board as the game evolves, and with this approach the AI player is able to mimic that ability.
Facebook is currently running a small test of a new AI assistant called M – a human-trained system which complete tasks on your behalf like shopping, orders and travel arrangements.
Check-out the video published here…