SETI Signal Detection

This repository contains a machine learning project aimed at detecting potential technosignatures of extraterrestrial intelligence in the form of anomalous signals. Data will be in the form of spectrograms provided by the Breakthrough Listen team at the University of California, Berkeley. The main goal of the project is to develop an algorithm that can identify these signals (referred to as "needles") from a dataset containing millions of signals, most of which are radio frequency interference (RFI) from human-made technology.

Dataset

The dataset consists of two-dimensional arrays, or spectrograms, representing the signal intensity as a function of frequency and time. These spectrograms are generated by the Green Bank Telescope (GBT), the largest steerable dish on the planet, with a diameter of 100 meters. The dataset is provided in the form of "cadence snippets," which are small regions of the spectrograms containing signals from target stars and other regions of the sky. The needles are simulated signals hidden within the haystack of data from the telescope.

Project Structure

project/
│
├── data/
│   ├── train.csv
│   ├── test.csv
│   └── train_labels.csv
│
├── notebooks/
│   └── exploratory_data_analysis.ipynb
│
├── src/
│   ├── models.py
│   ├── preprocessing.py
│   ├── train.py
│   └── utils.py
│
├── README.md
└── requirements.txt

Folders

• data: Contains the training and testing datasets in NumPy format.
• notebooks: Contains Jupyter Notebooks for exploratory data analysis and model experimentation.
• src: Contains the source code for the project, including model definitions, data preprocessing, and utility functions.

Files

• README.md: This file, containing an overview of the project and its structure.
• requirements.txt: A list of required Python packages for this project.

Getting Started

1. Clone the repository:

git clone https://github.com/yourusername/seti-signal-detection.git

2. Install the required packages:

cd seti-signal-detection
pip install -r requirements.txt

3. Explore the data using the provided Jupyter Notebook:

jupyter notebook notebooks/exploratory_data_analysis.ipynb

4. Develop and train your own models using the provided source code as a starting point:

python src/models.py

Data Visualization

Contributing

Please feel free to contribute to this project by submitting pull requests or opening issues with suggestions for improvements.

License

This project is licensed under the MIT License. See the LICENSE file for more information.