This repository contains my personal solutions for the "Reinforcement Learning" course at MIMUW. The course focuses on modern reinforcement learning techniques and algorithms, with emphasis on model-free methods, model-based methods, and search-based methods.
- Model-Free Methods
a) Markov Decision Processes (MDP) & Dynamic Programming (DP)
b) Value-Based Methods
c) Policy Gradient Methods
d) Actor-Critic Methods - Model-Based Methods
a) Model Estimation
b) Planning - Exploration
a) Multi-Armed Bandit Models
b) Search Strategies Related to Uncertainty - Research Topics
- Guest lectures from industry practitioners
The course aims to provide an understanding of the properties of reinforcement learning algorithms, develop skills to appropriately use methods for the development of dedicated reinforcement learning algorithms or apply existing methods in research projects, and enable students to implement their own algorithms and use existing libraries offering reinforcement learning procedures. It also fosters critical thinking, entrepreneurial action, and the importance of expert opinions.
The repository consists of Jupyter notebooks, scripts, and resources used to solve the course exercises and projects.
- Clone the repository
- Navigate into the repository
- Install required Python packages
- Launch Jupyter Notebook
This repository is for personal academic use. Plagiarism is not encouraged. If you're a student in the "Reinforcement Learning" course, refrain from copying or using this material for graded assignments. Use this for learning and understanding.