Convolutional neural network to perform binary classification by detecting patterns from the training dataset and making predictions on the test dataset.
Problem Statement: Lunar landings by renowned space stations across the world have yielded an abundance of new scientific data on the Moon. The various experiments placed on the surface provided information on seismic, gravitational, and other lunar characteristics. But perhaps the most dramatic result of the missions was returning a total of more than 800 pounds of lunar rock and soil for analysis on Earth. These samples of the Moon offered a deeper appreciation of the evolution of our nearest planetary neighbor.
Imagine you have been called by one of the largest space stations in the world (XYZ) space station and you are requested to make a Machine Learning model which classifies the different rocks present on the moon's surface. The purpose of this is to make the research process a lot easier. This will reduce the human effort of doing a monotonous task.
In this dataset, you will find 7534 images of 2 sizes of lunar rocks. In the next 2 months, we challenge you to build models such that given an image, the model will predict the probability of every rock class.
Approach: The problem has been solved using the Google Colabs platform. CNN has been used to extract features from the lunar rock sample images.
- The data set in drive has been accessed in colabs using pydrive module.
- Neural network -Activation layer -Relu of 128 density -2 softmax layers -loss function: categorical_crossentropy -optimizer: Adam
Dataset: https://drive.google.com/open?id=1ETMLnMFpLKCv4KT75CVZGKapYIVYCuUB