The DESI survey will measure large-scale structure using quasars as direct
tracers of dark matter in the redshift range $0.92.1$. We present two methods to select candidate
quasars for DESI based on imaging in three optical ($g, r, z$) and two infrared
($W1, W2$) bands. The first method uses traditional color cuts and the second
utilizes a machine-learning algorithm.