SDSS MLP Classifier

This project uses data from the Sloan Digital Sky Survey (SDSS) to classify celestial objects into three categories:

• Star
• Galaxy
• QSO (Quasar)

The classification is done using a Multi-Layer Perceptron (MLP) trained on both original and augmented data.

Folder structure:

• SDSS_DR18.csv: Original dataset
• augmented_SDSS.csv: Dataset after augmentation
• wrong_predictions.csv: Outputs where the model failed
• Preprocessing.ipynb: Prepares the dataset (cleaning, scaling, splitting)
• data_augmentation.ipynb: Adds synthetic or modified examples
• Multi-Layer Perceptron.ipynb: Trains the model and evaluates performance

Requirements:

• pandas
• scikit-learn
• matplotlib
• seaborn
• torch

You can install them with: pip install pandas scikit-learn matplotlib seaborn torch

To run the model, open Multi-Layer Perceptron.ipynb and run all cells.