Since humans could first postulate about outer space, our imaginations have been excited by the possibility of other life existing outside our own planet. In 2016 researchers at the University of Berkley began the Breakthrough Listen program to explore outer space for signs of alien life in the form of communications. Soon after, the Green Bank Observatory, Parkes Observatory, and Automated Planet Finder began detecting strange radio signals. Kaggle provides the data for these signals in the form of spectrograms in order to encourage an image classification-based approach. Using a combination of 1) Feature Engineering to develop unique insights about these signals and 2) Deep Learning with Convolutional Neural Networks, I aimed to build a robust, reusable model that accurately classifiies new images. Hopefully this data and this model lead to advancements that move us closer to being able to detect, and even communicate with, extraterrestrial beings in the future.
-
Data
-
Documents
-
Notebooks