To　deal　with　the　rapid　growth　of　mobile　data　traffic　and　the　user　interest　shift　from　peer-to-peer　communications　to　content　dissemination-based　services,　such　as　video　streaming,　information-centric　networking　has　emerged　as　a　promising　architecture　and　has　been　increasingly　used　for　wireless　and　mobile　networks.　In　this　paper,　we　focus　on　video　dissemination　in　information-centric　cognitive　radio　networks　(IC-CRNs)　and　investigate　the　use　of　harvested　bands　for　proactively　caching　video　contents　at　the　locations　close　to　the　interested　users　to　improve　the　performance　of　video　distribution.　With　consideration　of　the　dynamic　and　unobservable　nature　of　some　parameters,　we　formulate　the　allocation　of　harvested　bands　as　a　Markov　decision　process　with　hidden　and　dynamic　parameters　and　transform　it　into　a　partially　observable　Markov　decision　process　and　a　multi-armed　bandit　formulation.　Based　on　them,　we　develop　a　new　spectrum　management　mechanism,　which　maximizes　the　benefit　of　proactive　video　caching　as　well　as　the　efficiency　of　spectrum　utilization　in　the　IC-CRNs.　Extensive　simulation　results　demonstrate　the　significant　performance　improvement　of　the　proposed　scheme　for　video　streaming.