Skip to main content

The Use of Bayes’ Rule in Amazon GO Stores

This article attempts to figure out how Amazon Go supermarkets will work. These retail stores are special because they allow the shopper to grab the items they want then walk out of the store and go. This is a challenge because Amazon will have to determine the identity of the person who walks into the store and what items they have when they leave the store. The author of the article suggests this will be accomplished by a neural network that has been trained using massive amounts of data. For example, if you want your system to recognize soda cans, you show it tons of pictures of soda cans until it can recognize them.

This article is related to this course because the author suggests that Amazon is using Bayes’ rule to make their automated store work. For example, if the system could not identify an object and who picked it up, it would use an analysis like this: Given that the item was in these places in the store and that there are these other objects in the cart with the object, the object was likely picked up by a person who was in those places in the store and the object is probably something that person would buy. As the article states, this is similar to how NASA uses an algorithm based on Bayes’ rule called the Kalman Filter to account for the errors of multiple instruments to come up with one estimate.


Leave a Reply

Blogging Calendar

November 2017
« Oct   Dec »