Abstract: Low-cost gas sensors,with their advantages of compact size and affordable price,enable high-density grid deployment and have attracted extensive attention in the field of ambient air quality monitoring. Nevertheless,such sensors are susceptible to interferences including temperature variations, humidity fluctuations, as well as cross-interference from non-target gases in practical applications,leading to insufficient accuracy in measurement results. In recent years,the rapid advancement of machine learning technologies has provided a new technical pathway for improving the data quality of low-cost gas sensors,making machine learning-assisted low-cost gas sensing a current research hotspot. This paper reviews the classification and working principles of existing low-cost gas sensors,summarizes the current application status of machine learning technologies in this field,highlights representative research cases of performance enhancement for gas sensors assisted by machine learning, and prospects future development directions. It aims to offer references for relevant research in the fields of machine learning and gas sensing,thereby promoting technological innovation and progress in this domain.