In　this　paper,　we　propose　a　malware　identification　method　employed　by　image　analysis　and　generative　adversarial　networks,　designed　to　solve　the　problems　of　increasingly　sophisticated　attack　forms,　insufficient　sample　data　in　malware.　Specifically,　we　first　generate　fixed-size　gray　images　of　malware,　which　neither　disassembly　nor　code　execution　is　required　for　identification.　Moreover,　we　introduce　generative　adversarial　networks　into　malware　identification　for　few　samples　scenarios　and　malware　variants.　Through　the　game　training　of　generator　and　discriminator,　the　malware　detection　model　is　obtained　from　the　discriminator　and　the　samples　are　generated　by　the　generator　for　data　augment.　Finally,　we　demonstrate　that　the　proposed　method　is　efficient　and　feasible　using　extensive　experiments.　©　2021　IEEE.