Deep Q-Learning for Classification (Exploration Project)

This project explores the use of Deep Q-Learning for image classification task on the MNIST dataset.

---

Results

The following table summarizes the various metrics for the classification task:

Rewards and Epsilon During Training:

Evaluation Metrics on the trained model:

Metric	Value
Accuracy	95.72%
F1-Score	0.96
Precision	0.96
Recall	0.96

Heatmap for predicted vs true labels:

Running Project

To replicate the results of this project, follow these steps:

1. Set up a Virtual Environment

Run the following command to create a virtual environment:

python3 -m venv .venv

2. Install Poetry

Poetry is the dependency management tool used for this project. It simplifies package management, environment setup, and dependency tracking. Install Poetry by following the instructions on their official website.

3. Install Dependencies

Once Poetry is installed, use it to set up the project’s dependencies:

(.venv) poetry install

4. Run the Project

Execute the main script to start the training:

(.venv) python main.py