Resources for DRL and the actor-critic model submitted by students in Masters of Science in AI.
Forty-minute YouTube video on deep reinforcement learning and the basic concepts behind actor-critic models. This video is a very good introduction on the fundamentals of reinforcement learning.
Learn to implement an Advantage Actor Critic (A2C) agent that learns to play Sonic the Hedgehog.
Comprehensive with code examples - Easy interactive tutorial (runs effortlessly on Colab) - Familiar, TensorFlow style coding focus tutorial - hands-on guide
Easy format via relevant platform
Concise, plus video games
The Keras documentation offers a clear and concise example of implementing the Actor-Critic method in the CartPole environment.
This tutorial offers a step-by-step guide on implementing an Actor-Critic model.
Brief, Coding focus.
Explains the math behind Actor Critic architecture - demystify how Actor-Critic methods enable an agent (like a robot) to learn from interactions and feedback, improving its decision-making over time
If you want to take deep dive into deep reinforcement learning: Covers complex topic in very simple language and explanations.
In-depth exploration of Actor-Critic methods in the context of asynchronous reinforcement learning.
In-depth exploration of Actor-Critic methods in the context of asynchronous reinforcement learning
Insightful lecture series covering Actor-Critic methods and their applications in depth.
This is a 5 hour crash course training on the Advanced Actor Critic Methods.
To find comprehensive materials on a pivotal DRL approach that balances action decision-making (actor) with performance evaluation (critic) for learning optimization.
To explore specific implementations, variations, and advancements of Actor-Critic methods that show how theoretical models are applied in practical scenarios, enhancing machine learning capabilities.
This entire playlist is a very good learning resource for DRL. Lectures 7a and 7b gives an overview of the Actor and Critic model.
Explanation with math
Proposes a Deep Reinforcement learning based approach for Learning to rank task.
A lecture note from UC Berkeley's deep reinforcement learning course. It provides a theoretical understanding of Actor-Critic algorithms, making it a good resource for those who prefer the academic approach.
An approach based on the actor-critic framework, and in the critic branch we modify the manner of estimating Q-value by introducing the advantage function, such as dueling network, which can estimate the action-advantage value
This work views JSSP as a sequential decision making problem and proposes to use deep reinforcement learning to cope with this problem.