Advancements in Deep Reinforcement Learning: A Comparative Analysis of Policy Optimization Techniques
DOI:
https://doi.org/10.63282/3050-9246.IJETCSIT-V1I1P103Keywords:
Reinforcement Learning, Policy Optimization, Actor-Critic, Proximal Policy Optimization, Model-Based Learning, Deep Neural Networks, Sample Efficiency, Training Time, Advantage Estimation, Trust Region MethodsAbstract
Deep Reinforcement Learning (DRL) has emerged as a powerful framework for solving complex decision-making problems. This paper provides a comprehensive review and comparative analysis of various policy optimization techniques in DRL, including Policy Gradient Methods, Actor-Critic Algorithms, and Model-Based Approaches. We discuss the theoretical foundations, practical implementations, and recent advancements in each category. The paper also evaluates these techniques on a variety of benchmark tasks to highlight their strengths and limitations. Our analysis aims to provide insights into the current state of DRL and guide future research directions
Downloads
References
[1] https://fastercapital.com/content/Optimizing-Policies-in-Deep-Reinforcement-Learning.html
[2] http://www.jatit.org/volumes/Vol102No7/13Vol102No7.pdf
[4] https://www.mdpi.com/2226-4310/11/12/1040
[5] https://superagi.com/policy-optimization-algorithms-frameworks/
[6] https://jmlai.in/index.php/ijmlai/article/view/2
[7] https://www.researchgate.net/publication/332652066_Deep_Reinforcement_Learning_for_Optimization
orcement_Learning_for_Robotic_Manipulation_Towards_Trustworthy_Interpretable_and_Explainable_Artificial_Intelligenc
e
