Minecraft Diamond Challenge v Artificial Intelligence

Minecraft Diamond Challenge v Artificial Intelligence

it does not take long for most new players when starting Minecraft to work out how to excavate diamonds. These valuable blocks are a key feature to the game, however, trying to train artificial intelligence to do this task has proved easy, in fact, it has been more difficult than developers expected.

Last summer the developers at Microsoft who are now the publishers of Minecraft publisher and other organizations partnering with them on this project have found that artificial intelligence has really challenged coders trying to create AI agents that could find the coveted gems.

Some coders have been able to crack the problem with their first session but out of more than 660 entries submitted, not one was up to the task.

The aim of the competition had been to see if this problem could be solved without using a vast amount of computing power.

Although the results never provided any winners the organizers commented that they were still

“hugely impressed” by some of the participants.

“The task we posed is very hard,” said Katja Hofmann, a principal researcher at Microsoft Research. “Finding a diamond in Minecraft takes many steps – from cutting trees to making tools, to exploring caves and actually finding a diamond. 

“While no submitted agent has fully solved the task, they have made a lot of progress and learned to make many of the tools needed along the way.”

Mining Of Diamonds

As we know, Minecraft has become hugely popular since it was released in 2011 with more than 180 million copies of the open-world game sold globally along with the title that has more than 112 million monthly active players.

Diamond’s are one of the most important building resources in Minecraft as it is important in players being able to create strong armour and powerful weapons.

However, for a player to be able to get diamonds in Minecraft, this precious stone is only available when a player has completed a number of other steps.

“If you’re familiar with the game, it shouldn’t take more than 20 minutes to get your first diamonds,” Minecraft player Jules Portelli told the BBC.

Entrants to this competition were only allowed to use a single graphics when processing unit (GPU) and were permitted only four days of training time. Artificial Intelligence systems usually require many months or years of game time in order to master titles like StarCraft II.

A relatively small Minecraft dataset, with 60 million frames of recorded human player data, was also made available to entrants to train their systems.

MineRLImage copyrightMINERLImage caution participants had to submit their source code so that it could be verified by the organizers

Organizers of this project wanted the coders to create programs that were able to learn by example such as, through a technique that is known as “imitation learning”.

This method involves trying to get Artificial Intelligence agents to adopt the best approach by requiring them to mimic what humans or other software programs would use to solve a task.

It differs from relying solely on “reinforcement learning”, in which an agent is effectively being trained to find the best solution via the process of trial and error but without being able to draw on past knowledge.

artificial Intelligence

Researchers have now found that using reinforcement learning alone can sometimes deliver superior results. An example of this, for instance, is DeepMind’s AlphaGo Zero program which trumped one of the research hub’s earlier efforts, which in fact used both reinforcement learning and the study of labeled data from a human play to learn the board game Go.

this “pure” approach actually requires more computing power, This approach makes it too expensive for researchers to operate and only available large organizations or governments who have the financial budgets to do so.

William Guss, who was the main competition organizer and also a Ph.D. student at Carnegie Mellon University, explained to BBC that the point of the competition had been to show that:

“throwing massive compute at problems isn’t necessarily the right way for us to push the state of the art as a field”.

He added:

“It works directly against democratizing access to these reinforcement learning systems, and leaves the ability to train agents in complex environments to corporations with swathes of computing.”

The outcome, however, may have served to underline the advantage these well-funded entities actually have.

Minecraft is a game that is evolving all the time, it is a very versatile and engaging platform that captures children (and adult’s might I add) imaginations as well as stimulating school children with their education.

long may it continue!

Leave a Reply

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