Advertisement

A (cat) machine learning game I've been working on...

Started by May 11, 2019 01:37 AM
4 comments, last by jbarrios 5 years, 7 months ago

Wasn't sure where to post my game (https://incredicat.com) as I'm still working on it but wanted to put it out there to get any useful feedback or thoughts from the experts. It's basically a game similar to 20 Questions (or Animal, Vegetable, Mineral) that attempts to ask you questions to work out an object you are thinking about. You can think of everyday items (animals, household objects, food, quite a bit of other stuff etc) and it has 30 questions to try and guess the item. I've been working on it for a while but not sure what to do next so interested to hear anyone's thoughts...

I've finished working on the new algorithm which is based on ID3; entropy and information gain.

The link for anyone that wants to try it out is incredicat.com

Thanks in advance!

That's all folks!

Cool animation. I'm sorry, but how does entropy and information apply? I'm very interested.

Advertisement

Seems to need a lot more training. First it asked me if I was thinking of a type of cat (Yes), then if I was thinking about an animal (Yes - so not a mineral cat I guess) and THEN if I was thinking about a plant. ? So, one of the well-known plant-animals? ?

Something I wonder about : Why use the ID3 algorithm, given its well known drawbacks (easy overspecialization and tendency to get stuck on local optima)? Wouldn't the C4.5 or even C5 algorithms make more sense for a game like that? 

It got clock and cheeseburger right with about 18 guesses each.

🙂🙂🙂🙂🙂<←The tone posse, ready for action.

Hey Incredicat,

I played and recorded your game.  I recorded my thoughts in the video.
 

 

This topic is closed to new replies.

Advertisement