You can deploy this model and start earning money today!

AethronPhantom/
NexaAstro

Physics

Stellar Classification

tabular-classification

Allanatrix/Astro

apache-2.0

Views

No views yet

NexaAstro: Stellar Classification with SDSS Data

NexaAstro - Stellar Classification is a hybrid machine learning model for classifying stars using data from the Sloan Digital Sky Survey (SDSS). It leverages a two-stage architecture:

CatBoost: Gradient boosting for robust feature modeling.
Feedforward Neural Network (NN): Refines predictions for enhanced accuracy.

This model is part of the Nexa Scientific Model Suite, dedicated to advancing scientific discovery through machine learning.

Model Overview

Task: Multi-class stellar classification (e.g., Main Sequence, White Dwarf, Giant).
Input: SDSS stellar attributes (u, g, r, i, z magnitudes, spectral lines, etc.).
Output: Predicted stellar class label.
Architecture: CatBoost for feature extraction, followed by a Feedforward Neural Network for classification.

Applications

Stellar Population Studies: Analyzing distributions and characteristics of stellar types.
Galaxy Classification Support: Providing stellar data for broader galactic studies.
Astrophysics Education: Enabling interactive learning and research tools.
Feature Engineering: Supporting advanced astronomical machine learning workflows.

Getting Started

Example Usage

python

import joblib
import torch
import numpy as np
from my_nn_model import StellarNN  # Replace with actual neural network module

# Load CatBoost model
catboost_model = joblib.load("Allanatrix/catboost_model.pkl")

# Load PyTorch neural network model
nn_model = StellarNN()
nn_model.load_state_dict(torch.load("Allanatrix/stellar_nn.pt"))
nn_model.eval()

# Example prediction with SDSS features
features = np.array([...])  # SDSS input features (e.g., magnitudes, spectral data)
catboost_out = catboost_model.predict(features)
refined_pred = nn_model(torch.tensor(catboost_out).float())

Refer to the model documentation for detailed preprocessing and input requirements.

Dataset

Source: Sloan Digital Sky Survey (SDSS).
Preprocessing: Data cleaned, normalized, and filtered by magnitude thresholds.
Labels: Discrete stellar class labels derived from expert annotations.

Citation and License

If you use NexaAstro in your research, please cite this repository and acknowledge the SDSS dataset.
The model and associated code are licensed under the Boost Software License 1.1 (BSL-1.1).

Part of the Nexa Scientific Ecosystem

Explore related tools and models in the Nexa ecosystem:

Nexa Data Studio: Tools for data processing and visualization.
Nexa R&D: Research-focused model development environment.
Nexa Infrastructure: Scalable ML deployment solutions.
Nexa Hub: Central portal for Nexa resources.

Coming Soon:

Galaxy Morphology Classifier
Exoplanet Transit Detection Model

Developed and maintained by Allan, an independent machine learning researcher specializing in astrophysical and scientific AI systems.

AethronPhantom/NexaAstro