Search Unity

  1. Welcome to the Unity Forums! Please take the time to read our Code of Conduct to familiarize yourself with the forum rules and how to post constructively.
  2. We have updated the language to the Editor Terms based on feedback from our employees and community. Learn more.
    Dismiss Notice
New Forum User Notice Update to the Unity Editor Software Terms

ML Agents on Dynamic Difficulty Adjusting

Discussion in 'ML-Agents' started by TheJarmanitor, Aug 18, 2020.

  1. TheJarmanitor

    TheJarmanitor

    Joined:
    Mar 18, 2018
    Posts:
    20
    I'm not sure if it's correct to post here, but here we go.

    I'm interested on working my thesis with ML agents, speciffically on tuning the difficulty to make a game always engaging, using data from the player.

    I want to know if anyone has tried using unity and ml agents for Dynamic Difficulty Adjusting (DDA) and how were your experiences on the matter. Or if anyone can out me on a good direction.

    Thanks in advance.
     
    Last edited: Aug 23, 2020
    TheJarmanitor, Aug 18, 2020
    #1

  2. ervteng_unity

    ervteng_unity

    Unity Technologies

    Joined:
    Dec 6, 2018
    Posts:
    150
    The key here will be to come up with a reward function that rewards the agent (the "agent" could, for instance, have its actions be parameters of your game) for being "engaging", and giving the agent some observation about how the player is doing, e.g. their score.

    You'll also probably want to train/write a bot to mimic the player for training the difficulty agent. Pretty uncharted territory for me, but just my two cents!
     
    ervteng_unity, Aug 20, 2020
    #2

  3. TheJarmanitor

    TheJarmanitor

    Joined:
    Mar 18, 2018
    Posts:
    20
    thank you for answering! i'll take it into consideration. I am pretty scared to try because of the lack of information about it.
     
    TheJarmanitor, Aug 23, 2020
    #3