AI Talks: How Computers Make Decisions

jellybeans

Who said that artificial intelligence principles are too complex? Do you need a PhD to understand them? We don’t think so. In fact, we believe that you only need 2 things: some sweets and a little bit of clarity.

So, this week we will learn how computers are able to perform simple tasks such as deciding whether or not a picture contains a cat or a dog.

So, where is the cat and where is the dog? How can we make this decision automatically? What we need is a computer, a lot of cat and dog images and jelly beans.

 
puppersncates
 

Each one of them represents an image inside the memory of the computer. So, if we have multiple images, we will have multiple jelly beans. Then, we assign colours to each category: orange for dogs and green for cats.

 
Screenshot (206).png
 

Finally, we show the computer our images of dogs and cats.

 
.
 

Remember that in the memory of the computer, cat images are represented with green beans while dog images are represented with orange beans. The orange and the green beans represent the model for a cat and dog images.

 
Screenshot (211).png
 

We now want to make a decision: does this image contain a cat or a dog? We'll show a cat to the computer and then, the computer will compare the image with the preexisting models.

 
Screenshot (212).png
 

In this case, the cat model activates. What does this mean? It means that the computer decides that this image contains a cat.

 
Screenshot (216).png
 

So what did we learn in this episode? Firstly, that computers can recognise objects by looking at very distinctive details of such objects. Secondly, we learnt that computers make decisions by comparing what they see with what they have already seen in the past.

We hope you enjoyed this blog post! If you did, then don’t forget that we have the video version and more on our YouTube channel.