12. 準備:Branch and Bound Theoryって?
• Branch and bound is an algorithm design
paradigm for discrete and combinatorial
optimization problems, …
• The algorithm explores branches of this tree,
which represent subsets of the solution set.
• Before enumerating the candidate solutions of
a branch, the branch is checked against upper
and (lower) bounds on the optimal solution,
and is discarded if it cannot produce a better
solution than the best one found so far.
• The algorithm depends on the efficient
estimation of the (lower) and upper bounds
of a region/branch of the search space
[Wikipediaから一部抜粋&加筆]
13. 準備: BnBの例 ナップサック問題
• 問題:容量100KgのバッグにValueが最大
になるようにItemを詰めたい
• 条件:Itemは分割NG→離散最適問題
A BRANCH AND BOUND ALGORITHM FOR THE KNAPSACK PROBLEM*t
PETER J. KOLESAR (Peter J 1967)
容量100Kgのナップサック
ItemのWeightとValue
愚直にやると7つのItemを入れる/入れないで
27通り試す必要あり
※ナップサック問題はUpper Boundのみを計算
21. 参考資料
• Steven Gold+ "New algorithms for 2D and 3D point matching:
pose estimation and correspondence," Pattern Recognition, Vol.
31, No. 8, pp. 1019-1031, 1998
• PETER J +, ”A BRANCH AND BOUND ALGORITHM FOR THE
KNAPSACK PROBLEM”
• Yang+,”Go-ICP: A Globally Optimal Solution to 3D ICP Point-Set
Registration”
• Makadia, A+ “Fully Automatic Registration of 3D Point Clouds”
• Nicolas+ “Super4PCS: Fast Global Pointcloud Registration via
Smart Indexing”
• Drost+ “Model globally, match locally:Efficient and robust 3D
object recognition”
• ICP(Wikipedia)