[PYTHON] I tried to make Othello AI with tensorflow without understanding the theory of machine learning ~ Introduction ~
Series table of contents
--I tried to make Othello AI with tensorflow without understanding the theory of machine learning-Introduction- -I tried to make Othello AI with tensorflow without understanding the theory of machine learning-Implementation- -I tried to make Othello AI with tensorflow without understanding the theory of machine learning-Iza battle- -I tried to make Othello AI after trying to understand the theory of machine learning ~ Restart! ~ -I tried to make Othello AI after trying to understand the theory of machine learning-What is this Alpha Zero edition- -I tried to make a neutral network with Excel to understand the theory of machine learning ~ Image recognition mnist edition ~
In this field, as an outsider, I didn't study "theory of machine learning" at all. I would like to make an AI for Othello. Click here for the referenced site ・ Implement DQN (Deep Q Network) with TensorFlow in a super-simple manner ~ Introduction ~
Overview
If you explain the concept roughly
Two artificial intelligences confront Othello earnestly,
Save the AI behind and fight me (human). That's why.
Environmental condition
My environment is as follows. OS ・ Ubuntu Development environment ・ Python 3.5
I think it works in this environment Building Ubuntu python development environment on Google Cloud Platform
Let's move it first
First, download the source code.
The source is here.
$ git clone https://github.com/sasaco/tf-dqn-reversi.git
Learning
When the environment is ready, move to the source code directory and hit train.py to start learning.
python
$ cd tf-dqn-reversi
$ python train.py
If you see the following log, you are learning correctly.
python
player:1 | pos:32 | LOSS: 0.0000 | Q_MAX: 0.0041
player:2 | pos:15 | LOSS: 0.0000 | Q_MAX: 0.0009
…
layer:2 is only pos:56
player:2 | pos:56 | LOSS: 0.0000 | Q_MAX: 0.8607
EPOCH: 999/999 | WIN: player2
winner is player2
- LOSS is almost 0, but is there any problem? [.](Http://ja.stackoverflow.com/questions/30312/tensorflow%e3%81%a7dqn-%e3%81%aa%e3%81% 9c% e3% 81% 8bq% e6% 9c% 80% e5% a4% a7% e5% 80% a4% e3% 81% 8c% e5% b0% 8f% e3% 81% 95% e3% 81% 84) I don't even understand what LOSS is (￣ ^ ￣) If you are familiar with this, please let us know. m(_ _)m For implementation, see Implementation.
test
It will take several hours to finish the study. Now let's test with the model we learned.
python
$ python FightWithAI.py
------------- GAME START ---------------
***user turn ○***
0 1 2 3 4 5 6 7
8 9 10 11 12 13 14 15
16 17 18 19 20 21 22 23
24 25 26 ○ ● 29 30 31
32 33 34 ● ○ 37 38 39
40 41 42 43 44 45 46 47
48 49 50 51 52 53 54 55
56 57 58 59 60 61 62 63
Please enter the number
[43, 34, 29, 20]
>>>
If the game starts as above, you are successful. Did it work properly? The result of the match will be written in ~ Iza Battle ~.
Next time will deliver the implementation version.
Recommended Posts