Thispapertacklestheefficiencyproblemofmakingrecommendationsinthecontextoflargeuseranditemspaces.Inparticular,weaddresstheproblemoflearningbinarycodesforcollaborativefiltering,whichenablesustoefficientlymakerecommendationswithtimecomplexitythatisindependentofthetotalnumberofitems.Wep