Learn More About Facebook AI Research



if you think about ambitious sensory questions over times there is you know what's the universe made of what life is all about and how the brain works and you know AI basically belongs to that third category artificial intelligence is a set of techniques that allows machine to do tasks that are normally attributed to animals and humans the history of this is very personal for me so I came up with this idea of combinational net which was a way of organizing the artificial neurons in the neural net in such a way that it could be applied to things like image recognition at that time we didn't realize how much of an impact this could have but by the mid 90s you could go to an ATM machine and put the check in the ATM machine an ATM machine would read the amount on the check automatically that was using a conditional neuron act that we developed at Bell Labs you could think of a control net as composed of very simplified simulated neurons brain cells that are connected with each other and the way this comes on that works is by adjusting the strength of the connections between those neurons the same way the brain does it so this is a real-time demo of object recognition it's a combinational neural network but it runs on a laptop in real time and ten times per second we can tell you what object is in the image neuroscientists tell us that a pretty big part of our visual system is actually just devoted to recognizing faces so facial recognition is a subfield of computer vision essentially so it's a particularly interesting problem within computer vision and it's special because there is nothing that looked more like like each other than two faces and there are a lot of people that look like each other so recognizing faces discriminating faces from each other is very very difficult we show two faces of the same person perhaps at different times or different poses or wearing glasses no glasses etc and we tell the system whatever output you produce they should be similar for those two images because there are images of the same person and then we show also pairs of images of different people and we tell the system produce different outputs for them and then we can figure out how similar are two objects according to the machine so we can run lots and lots of images to the machine and then try to lay out all those two images similar or not if they are similar we're going to put them next to each other and if they are not similar we're going to put them far away from each other here at Facebook our research we publish a lot research we publish most of the code that we write in open source so that we can interact with the rest of the research community to solve this very challenging problem of building intelligent machines or making machines more intelligent intelligence is one of the most mysterious things around that's what makes us uniquely human is intelligence if you want to understand ourselves we have to understand what intelligence is all about

7 thoughts on “Learn More About Facebook AI Research”

  1. I would love to have a chat with you the reason why is I am releasing software that already does what you are wanting to do but I have already done it… here is link to youtube video showing it I have more on software and we are underway to finishing website as soon as we secure funding to launch. https://www.youtube.com/watch?v=T13B-wsGB4Q

Leave a Reply

Your email address will not be published. Required fields are marked *