I suggest two possible approaches. Narasaka, et al. have effected similar cyclizations photochemically. I prefer to start with the allylic alcohol 1 to avoid complications with oxime formation. Formation of the bis acetate 2b is not a problem. However, given the greater acidity of oximes relative to alcohols does not preclude the formation of oxime acetate 2a. In the photolysis the naphthalene serves as a sensitizer while 1,4-cyclohexadiene functions as a hydrogen atom donor. The addition of hydrogen is likely from the convex face as shown. If not, after saponification of acetate 3b and Swern oxidation, base equilibration would likely lead to ketone 4.
Alternatively, a Barton-type radical cyclization (5 --> 3a) might be considered.
ADDENDUM: Direct photolysis of a structure similar to 5 without Bu3SnH has been achieved. See ref 2.
1) M. Kitamura, Y. Mori and K Narasaka, Tetrahedron Letters, 2005, 46, 2373.
2) F. Gagosz and S. Z. Zard, Synlett., 1999, 1978.