Gas classification in a wind tunnel using 1D CNN PyTorch
This work aims to familiarize with the process, dataset, and corresponding documentation. The code is written in Python and the library for machine learning is PyTorch. This code still has quite low accuracy in classifying various gasses in the dataset and still has plenty of room for improvement.
For the dataset see: Chemical Detection Platform
The dataset contains 18000 time-series recordings from a chemical detection platform at six different locations in a wind tunnel facility in response to ten high-priority chemical gaseous substances. The resulting dataset induces a ten-class gas discrimination problem.