Development of algorithms for multiple and robust solution search and Their Applications to Knowledge-Base Optimization in Japanese-Braille Translation

本研究提出的算法，搜索有前途的，准最优的解决方案，除了最优的解决方案，并搜索解决方案的鲁棒性对设计变量的小扰动。本研究还表明，在日语盲文翻译知识库建设中的定义相似性度量的问题可以制定为一个优化问题，所提出的算法可以应用于这个问题和其他问题。

This research proposed algorithms which search for promising, quasi-optimal solutions in addition to optimal solutions, and which search for solutions robust against small perturbations of design variables. This research also showed that a problem of defining similarity metric in knowledge base construction for Japanese-Braille translation could be formulated as an optimization problem, and that the proposed algorithms could be applied to this problem and other problems.