Scribd
Upload
3 views

Final Project

The final project aims to summarize the semester's learning, review reinforcement learning algorithms, and understand their practical applications. Students are required to choose a topic related to reinforcement learning, investigate it, and discuss various aspects such as problem-solving, mathematical principles, and algorithm implementation. The project must be submitted by January 27, 2024, in a specified format including a PDF and code with a README file.

Uploaded by

Syed Wajahat
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
3 views

Final Project

The final project aims to summarize the semester's learning, review reinforcement learning algorithms, and understand their practical applications. Students are required to choose a topic related to reinforcement learning, investigate it, and discuss various aspects such as problem-solving, mathematical principles, and algorithm implementation. The project must be submitted by January 27, 2024, in a specified format including a PDF and code with a README file.

Uploaded by

Syed Wajahat
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 5

Final Project

TA- 张君瑞 [email protected]


Purpose

• Summarize the learning content of this semester.

• Review the development of reinforcement learning algorithms.

• Gain a basic understanding of the practical applications of reinforcement


learning algorithms.
Content
This experiment is primarily focused on research. It is recommended that you choose one of the following topics for

your investigation and related theoretical studies and discussions.


• Reinforcement Learning Algorithm
A. TD3
B. TRPO
C. PPO
D. DDPG
E. SAC
• Extended(Bonus)
F. Imitation Learning
G. MARL(Multi Agent Reinforcement Learning)
H. Hierarchical RL
I. Offline RL
J. Multi-task RL
K. Safe RL
L. Application of RL
Requirements
• The research can be discussed in the following aspects ：
1. What kind of problems does this field aim to solve?
2. Mathematical principles (including the mathematical modeling of the problem)
3. What are the classic algorithms in this field? What are their strengths and weaknesses? How
can they be improved?
4. Implement at least one algorithm, present the training results, and provide an analysis.
The above aspects are suggestions and not exhaustive.
• Notes ：
1. Necessary description of the implemented code, including but not limited to: environment
preprocessing, action space, state space, rewards, main parts of the code ......
2. A~E ： You can focus on the improvements of the algorithm and compare the improved
algorithm with the original one.
3. L ： You can choose a practical problem related to your research field or of personal interest,
model it as an RL problem, and solve it using an appropriate RL algorithm.
4. The chosen topic is not limited to A~L mentioned earlier but must be in the RL-related domain
and should follow the above requirements.
Submission
1. Submit to ： bb.ustc.edu.cn
2. Deadline ： 2024-1-27 23:59:59
3. Format: student_name_exp4.zip
4. Content: pdf+code (please include a separate README file to explain
how to run the code)

You might also like