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

Multi criteria/multi objective in the reward: adaptative weight ?

Discussion in 'ML-Agents' started by ucax217, May 14, 2022.

  1. ucax217


    Nov 8, 2018

    I a an agent with several term/criteria/objective in the computation of the reward.

    I have setup a curriculum learning by changing the weight of each criteria. At first the weights are set to focus on the easy and essential things, then the weights change to give importance to the details that will gain quality.
    It works a little but not enough.

    I would like to know if there is algorithm in Ml-Agent to automatically adapte the weight in multi-objective reward.

    I have found some papers like this :
    Prediction-Guided Multi-Objective Reinforcement Learning for Continuous Robot Control
    Jie Xu, Yunsheng Tian, Pingchuan Ma, Daniela Rus, Shinjiro Sueda, Wojciech Matusik
    International Conference on Machine Learning (ICML) 2020

    Thank you