Monitoring Air Quality using Domain Adaptation and GAN

Abstract

Recent environmental pollution is most detrimental to air which is one of most necessary elements of nature for living beings. Recent excessive ozone pollution made the Texas Commission on Environmental Quality (TCEQ) to call for an Ozone Action Day for the Austin area for Saturday, March 19, 2022 [1] to prevent ozone pollution. But in most of the cases predicting air quality with the help of data collected from somewhere else doesn’t help much. Moreover measuring directly the amount of pollution may not be feasible always. Collecting directly the elements to predict one specific amount also may not be feasible. So, in this work it has been tried to measure one element from other available elements without taking account their contribution to the specific element. Here measuring ozone from other data has been tried out. Also taking elements from one city air it has been experimented to measure pollution in a complete different city. By studying this real life scenario it has been tried to measure the performance among different domain adaptation methods including optimal DANN and optimal transport and subspace alignment to measure their effectiveness in this real life scenario.