To explicitly understand the spatio-temporal pattern change of desertification and its driving mechanism in China-Mongolia-Russia economic corridor proposed by the "Belt and Road Initiative", the classification and regression tree (CART), support vector machine (SVM), random forest (RF) and Albedo-NDVI model were compared to better monitor desertification on Google Earth Engine platform. Furthermore, the climate and human factors were introduced into the linear trend method to quantitatively measure the driving effect of each factor on the desertification process. The results show that 1) the classification regression tree model combined with multi-source data can better monitor desertification in the study area. The overall accuracy and kappa coefficient are 85% and 0.754, respectively. 2) From 2000 to 2015, although the area of land degradation and strong degradation was 17,800 km² more than the area recovered and significantly restored, the area of extremely severe, severe and moderate desertification was decreased by 73,100 km², 43,200 km² and 13,900 km², respectively. In addition, the desertification area in China has recovered significantly, with a net restoration area of 151,800 km². The desertification in Mongolia and Russia is worsening. 3) During the study period, climate change not only played a major role in desertification restoration in the study area, but also played a major role in desertification restoration in various countries. The regions with the main driving force of climate change, accounted for 68.8% of the total desert restoration areas in the study area. In areas where desertification is increasing, human activity is driving more than climate change, accounted for 69.68% of the total desert restoration areas in the China-Mongolia-Russia economic corridor.