# AMLS_24-25_SN12345678

This project implements machine learning models for the **BreastMNIST** and **BloodMNIST** datasets, tackling two key tasks:

1. **Task A: BreastMNIST Binary Classification**  
   Classify images as "Benign" or "Malignant."

2. **Task B: BloodMNIST Multi-Class Classification**  
   Classify images into one of eight blood cell types.

The project is structured into folders for each task, as well as supporting datasets and utilities.

---

## Project Organization

### 1. **Folder A: Task A - BreastMNIST Binary Classification**
This folder contains the implementations for Logistic Regression and Support Vector Machines (SVM) models:

- **`A_logistic_regression.py`**  
  Logistic Regression implementation using separate training and validation datasets.

- **`A_logistic_regression_combined_dataset.py`**  
  Logistic Regression implementation using combined training and validation datasets for model training.

- **`svm.py`**  
  SVM implementation using separate training and validation datasets.

- **`svm_combined_dataset.py`**  
  SVM implementation using combined training and validation datasets for model training.

### 2. **Folder B: Task B - BloodMNIST Multi-Class Classification**
This folder contains CNN-based implementations for the multi-class classification task:

- **`CNN.py`**  
  Basic CNN implementation for BloodMNIST classification.

- **`CNN_tuning.py`**  
  CNN implementation with hyperparameter tuning using Optuna for optimized performance.

- **`pretrained_CNN.py`**  
  Implementation of a pre-trained CNN (e.g., ResNet18) for BloodMNIST classification using transfer learning.

- **`pretrained_CNN_tuning.py`**  
  Implementation of a pre-trained CNN with Optuna-based hyperparameter tuning. This script finds the best hyperparameters but does not evaluate the model.

### 3. **Datasets**
- **`BreastMNIST`**  
  Dataset for binary classification of breast tumor images.

- **`BloodMNIST`**  
  Dataset for multi-class classification of blood cell types.

### 4. **Main Files**
- **`main.py`**  
  A central script that integrates the various models for both tasks. This can be used to run the entire project pipeline.

- **`.gitignore`**  
  Specifies files and directories to be ignored by version control.

- **`requirements.txt`**  
  Lists all required Python packages for the project. Use this file to install dependencies with `pip install -r requirements.txt`.

---



## Required Packages

The following packages are used in this project. Make sure to install them before running the scripts. All required packages are also listed in `requirements.txt` for easy installation.

### Python Libraries

- **NumPy**: Numerical computations and array manipulations (`numpy`)
- **scikit-learn**: Machine learning tools and metrics, including:
  - `LogisticRegression`
  - `SVC`
  - `GridSearchCV`
  - `StandardScaler`
  - `classification_report`
  - `confusion_matrix`
  - `accuracy_score`
  - `learning_curve`
- **Matplotlib**: Visualization of learning curves and results (`matplotlib`)
- **Torch**: Deep learning framework for building and training neural networks (`torch`)
  - `torch.nn`: For defining neural network layers
  - `torch.optim`: Optimization algorithms
  - `torch.utils.data`: Data loading utilities
  - `torchvision.transforms`: Image transformations
- **Optuna**: Hyperparameter optimization library for tuning models (`optuna`)
- **Torchvision**: Pre-trained models and additional utilities for deep learning tasks (`torchvision`)

---

### Installation

To install all required packages, run:
```bash
pip install -r requirements.txt

To run different scripts, you can replace the script path in your Python code. For example, to run the `A_logistic_regression_combined_dataset.py` script located in the `A` folder, you can use the following code snippet:

```python
import os

# Replace the path to run different scripts
a_script_path = os.path.join(os.path.dirname(__file__), "A", "A_logistic_regression_combined_dataset.py")