With　the　aging　of　the　population　gradually　serious　in　recent　years,　the　research　of　multi-resident　activity　recognition　in　smart　home　has　been　paid　much　attention.　For　this　reason,　an　advanced　rule-based　expert　system,　called　cumulative　belief　rule-based　expert　system,　is　introduced　to　develop　a　novel　multi-resident　activity　recognition　model,　which　not　only　makes　full　use　of　the　multiple　labels　of　residents＇　activities,　but　also　can　overcome　the　problem　of　excessive　data　collected　from　smart　home.　In　the　case　study,　the　experimental　study　shows　that　the　proposed　model　is　more　efficient　and　accurate　than　the　traditional　machine　learning　models　and　the　commonly　used　activity　recognition　model　for　achieving　multi-resident　activity　recognition.　©　2023　IEEE.