Inspired　by　the　orthogonal　time　frequency　space　modulation,　carrier　index　differential　chaos　shift　keying　(CI-DCSK)　system　has　been　proposed　and　attracted　more　and　more　attention　because　of　its　high-efficiency　and　low-complexity　advantages.　In　this　paper,　a　deep　learning　based　intelligent　detector　for　CI-DCSK　system,　referred　to　as　DL-CI-DCSK　detector,　is　proposed　to　realize　more　reliable　transmission.　The　proposed　detector　inherits　the　advantages　of　neural　network　and　traditional　energy　detection.　The　proposed　DL-CI-DCSK　detector　first　recovers　the　index　bits　using　a　neural　network　and　then　using　the　index　bits　to　demodulate　the　modulation　bits.　The　designed　network　structure　mainly　exploits　the　characteristics　of　long　short-term　memory　unit　and　multiple　fully　connected　layers　to　extract　and　integrate　the　correlation　and　features　of　the　modulated　signals.　Simulation　results　show　that　the　proposed　intelligent　detector　can　achieve　better　bit-error-rate　(BER)　performance　than　conventional　detectors　over　multipath　Rayleigh　fading　channels.　©　2023　IEEE.