sure, you can change it! If you use Maxent from R (package 'dismo'), then it's easy: you should change the parameter 'ngb' of function maxent() to 20000 from the default value (10000). According to the help file, these are sampled randomly from the cells that are not NA in the first predictor variable.
But, my question means if I change Max number of background points to 20000; My analyses will be correct or no? Otherwise I know to chage this value to 20000
According to Phillips et al. 2009, background points allow you to characterize the environments in the study region. In other words, It establishes the environment domain of the study while the presence data establish under which conditions the species is more likely to occur in average.
As such, increasing the background points will surely improve the models outputs.
Ideally, one can use all the cells of study region without NA as background points but the computation time could increase for a very large region.