Network　is　an　important　representation　method　to　describe　related　objects.　For　network,　the　most　core　research　is　to　reasonably　represent　the　characteristic　information　of　nodes　in　the　network,　which　is　called　network　representation　learning.　In　recent　years,　many　scholars　have　proposed　many　excellent　representation　learning　algorithms,　through　which　the　original　network　data　can　be　embedded　in　low-dimensional　representation,　which　can　help　us　classify　the　nodes　in　the　network,　and　the　nodes　can　also　be　used　as　point　coordinates　in　Euclidean　space　for　visualization.　Existing　algorithms　are　all　aimed　at　embedding　the　original　data,　but　how　to　use　the　embedded　data　to　restore　the　original　data　when　the　original　data　is　incomplete　has　not　been　studied　by　scholars.　In　order　to　solve　this　problem,　this　paper　proposes　two　solutions　of　deep　learning,　one　is　the　artificial　neural　network　method　based　on　deep　learning,　and　the　other　is　the　attention　mechanism　method　based　on　deep　learning.　The　experimental　results　of　this　paper　show　that　these　two　methods　are　very　effective.　©　2019　IEEE.