rene_mobile’s avatarrene_mobile’s Twitter Archive—№ 1,925

  1. #CCS16 dl.acm.org/citation.cfm?id=2978392&CFID=686080847&CFTOKEN=72761292 tricking deep learning face recog. with printed glasses to masquerade as somebody else or for invisibility
    1. …in reply to @rene_mobile
      This specific example underlines what we knew for long: neural network black boxes are hard to debug and thus hard to trust for security