Entropy Bonus

مكافأة العشوائية الدلالية

حدّ يُضاف لدالة الهدف يكافئ السياسة على الاحتفاظ بتوزيع احتمالي منتشر بدل التركيز على فعل واحد، مما يشجع الاستكشاف ويمنع الانهيار المبكر.

A term added to the objective that rewards the policy for maintaining a spread-out probability distribution rather than concentrating on one action, encouraging exploration and preventing premature convergence.

Also translated asحافز الإنتروبيا، مكافأة التنوع

First appears in this corpus in: Asynchronous Methods for Deep Reinforcement Learning (2016)

Appears in these papers

Asynchronous Methods for Deep Reinforcement Learning2016 in the sky ✦
Asynchronous Methods for Deep Reinforcement Learning2016 in the sky ✦
First Return, Then Explore2021 in the sky ✦
Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor2018 in the sky ✦
Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor2018 in the sky ✦