This website uses cookies and similar technologies to understand visitors' experiences. By continuing to use this website, you accept our use of cookies and similar technologies,Terms of Use, and Privacy Policy.

Jul 31 2017 - 08:00 PM
Learning to Think Like a Computer

Anytime you ask Siri a question or tag someone in a photo, a neural network is being activated. These networks or neural nets are loosely inspired by the brain. These connections of artificial neurons mimic real neurons by triggering in a cascading way down a change of neurons until one result is reached. Neural networks differ from human brains by using a trick called back propagation by sending information back through the network so that the neural network as a whole can improve. Neural networks become black boxes due to their many complex layers.

To better understand these complex systems, researchers are taking different approaches to learn the inner workings of neural networks. Though we may never understand growing neural networks, researchers have tried to learn slivers of information. They are using unusual techniques like teaching an AI to play the game Frogger as a human would to learn more about the peaks and valleys of neural networks.

How could neural networks save time or increase human capacity? In what ways? Tell us about the possibilities of neural networks in the ongoing discussion on Vialogues.

Excerpts from the discussion

@00:57 csd2126: I can see how neural networks save time by automating portions of complex processes and allowing human judgment to make the final call. Instead of scrolling through a list of everyone I know, a computer system is able to take a guess at who might be in one of my pictures so I can agree to tag them in it.

@01:34 Claude.Bosse: I see that there is a danger in relying entirely on the judgments of computers as errors or prejudice may be built into systems which are used to make decisions affecting people's lives.

|By: Caitlin Davey|650 Reads