Skip to content

Automatic Detection of Martian bow shock crossings using data of the Mars Express mission: A Deep Learning Approach

Notifications You must be signed in to change notification settings

menouarazib/martian-shock-detection

Repository files navigation

Martian Bow Shock Detection

Automatic Detection of Martian bow shock crossings using data of the Mars Express mission: A Deep Learning Approach to see more chek this link : https://pnst-2022.sciencesconf.org/399962.

Abstract

We investigate to automatically detect the Martian bow shock crossings using the data of the Mars Express mission provided by CDPP-AMDA. Using a Multilayer Perceptron Neural Network, we provide an automatic classifier to predict the Martian bow shock crossings. A published catalog with around 11800 bow shock crossings has been used for labeling the data [1]. The challenging task was to deal with the unbalanced data, indeed, in our dataset, we have unequal distribution of classes: shocks and no shocks. Classification of unbalanced data is a difficult task because there are so few samples (shocks) to learn from. To tackle this problem is to penalize the mis-classification made by the minority class by setting a higher class weight and at the same time reducing weight for the majority class.

[1] B. E. S. Hall et al. “Annual variations in the Martian bow shock location as observed by the Mars Express mission”. In: Journal of Geophysical Research: Space Physics 121.11 (2016), pp. 11, 474–11, 494

Metrics

Metrics Definition

alt text

Model's Metrics

Is this section you will find out the values of different metrics obtained during a training phase over one month of data (07-2012):

alt text

Tests

Example: python -m run_predictions_orchestra . --start 2008-07-03T00:00:00 --stop 2008-07-05T00:00:00

alt text

Zoom in between 2008-07-04T05:27:34 and 2008-07-04T08:40:20 => alt text

Description

This repository contains a python script (run_predictions_orchestra.py) which should be executed in a virtual environment in order to make prediction of Martian Bow Shock crossings associated with a given start and stop times.

How this works ?

To do predictions you need to follow these steps: 👏
  • you should have python >= 3.10 already installed
  • activate the virtual environment = venv\Scripts\activate
  • install the requirements = pip --no-cache-dir install -r path/../requirements.txt
  • run the script with the following arguments a destination folder path to store results, a start time start and a stop time stop => python -m run_predictions_orchestra . --start 2008-07-03 --stop 2008-07-06

👏

About

Automatic Detection of Martian bow shock crossings using data of the Mars Express mission: A Deep Learning Approach

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages