Mars Spectrometry 2: Gas Chromatography -- Second place solution
This is an incremental solution for a specific NASA competition on analyzing gas chromatography data from Mars.
The paper tackled the Mars Spectrometry 2: Gas Chromatography challenge by using 2D image-like representations of chromatography data and ensembling Convolutional Neural Networks, achieving the second-best score on the test dataset.
The Mars Spectrometry 2: Gas Chromatography challenge was sponsored by NASA and run on the DrivenData competition platform in 2022. This report describes the solution which achieved the second-best score on the competition's test dataset. The solution utilized two-dimensional, image-like representations of the competition's chromatography data samples. A number of different Convolutional Neural Network models were trained and ensembled for the final submission.