Efficient computation of robust low-rank matrix approximations in the presence of missing data using the L1 norm | IEEE Conference Publication | IEEE Xplore