Machine Learning: How to Learn an MTG Card Game

Question

I became acquainted with many different approaches to computer training, but I had problems determining which approach might be the most suitable for my given funny problem. (IE: this is a controlled learning problem, and if so, what is my input x and output y?)

The magic of a compilation project consists of many actors sitting around a table holding a pack of 15 cards. players select a card and transfer the remainder of the packet to the player next to them. They open a new package and do it again in 3 rounds (45 decisions). People end up with a deck that they use to compete.

I am having trouble understanding how to structure the data that I have in the trials that are needed for training. I want a solution that 1) builds knowledge about which cards are selected compared to previous cards that are selected 2) can then be used to decide which card to choose from this new package.

I have a human choice data set that I would like to know about. It also includes information about the cards that they ultimately collected, but ultimately discarded. What could be an elegant way to structure this data for training, aka, what are my functions?

Answer 1

, markov. , / . - - ( , ..). , :

• Q-
• Sarsa
• UCT

, " : " .

Answer 2

, , - - . . , - : , ( AI), , .

, , ... . , , , , , " X 8".

, , . , , .

. , , - "", .

, (.. ) , , , . , 6 , RHO - , - RHO 5 6 , ( ) .

... . - , . , ( , ..). .

, . , : , , , .

Answer 3

1. "" - , A: ( ) B: ( ).

2. , , , " ", A + B + ( ).

3. , 1 2 - 1 "" ( ) 13 "" ( ), 14 .

4. , . ( , , .)

5. .

6. , , (), A + B +, , . ( , . , , .)

7. , . , .http://www.cs.toronto.edu/~mvolkovs/recsys2018_challenge.pdf