To increase the complexity of a $k-$nearest neighbors $($kNN$)$ model, we can

lower $k,$ the number of nearest neighbors

decrease the size of the training sample

increase the size of the training sample

increase $k,$ the number of nearest neighbors