a MACHINE of LEARNING : Machine Learning applied to Game of Thrones

Art

As the final season of Game of Thrones continues, this is a timely glance at the application of Machine Learning (ML) and Artificial Intelligence (AI) to both the books and the plots of the TV shows. Be warned, as mentioned in the title, this article and links within may / will  contain spoilers!

Does Data know who dies ?

A team at the Technical University of Munich have developed a machine learning algorithm intended to predict the likelihood of survival for the main characters in Game of Thrones. Using Longevity Analysis applied to a variety of character traits and events, a percentage figure is arrived at, suggesting the likelihood of death, and potentially therefore a seat on the Iron Throne. Full details, as well as many other insights gleaned from analysis of data can be found here: A Song of Ice and Data. Though the predictions are being updated as characters are killed off, so be wary of finding out fatalities before watching the show.

What might be next ?

Readers of the books and/or watchers of the series will be aware that the TV show finished using the books as source material after the 5th book, electing to continue the plot before George R R Martin had published the sixth and seventh books. This hiatus in the story prompted Zach Thoutt, an IT engineer to train a recurrent neural network to ‘write’ the next chapters. Several interesting predictions came from this process, yet to be realised in either book or TV formats, but potentially close to what is to come. Five of these chapters can be accessed here: Neural Network generated chapters.

Artwork Explained

In conclusion, there is an explanation for the picture used as the main image for this article.

It was generated using Ganbreeder ; an app that generates images based on combining and breeding images to produce a variety of offspring from the originals. Based on the limited number of parent images available, the 2 parent photos were of a dragonfly and a throne; the closest approximation to a dragon and throne combination! The results posted above are 3rd and 4th generation iterations of the combination of a dragonfly and a throne. This is, admittedly, a somewhat tenuous link, but one that delivered interesting results.

Sources:

https://www.valuewalk.com/2019/04/machine-learning-who-will-survive-game-of-thrones/

https://got.show/

https://medium.com/h-farm-innovation/artificial-intelligence-is-writing-the-new-book-of-game-of-thrones-9c3043f616e8

https://github.com/zackthoutt/got-book-6/tree/master/generated-book-v1

https://ganbreeder.app/

 

 

 

Summary
a MACHINE of LEARNING : Machine Learning applied to Game of Thrones
Article Name
a MACHINE of LEARNING : Machine Learning applied to Game of Thrones
Description
Brief overview of machine learning applications using data from Game of Thrones.
Author
Publisher Name
mlreporter.com
Publisher Logo
No Comments

Leave a Reply

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

Art
ARTificial Intelligence & Machine Learning in the Arts

  Art generated by Artificial Intelligence (AI), or more accurately Machine Learning (ML) is increasingly making not only technology news, but also being reported in the mainstream mass media. While computer generated art is by no means a new phenomenon, its increased exposure throws up both opportunities and challenges in …