Near optimal solutions navigator
Michael E. Smoot, William R. Pearson, Stephanie A. E. Guerlain · 2003
Sometimes, "optimal" solutions fail to meet operational objectives. Nevertheless, we expect that the operationally optimal solution lies near the algorithmically optimal solution. Various algorithms can be used to generate near optimal, alternative solutions, but there are significant challenges In displaying the multiple solutions to the decision maker, and to provide the additional Information necessary for the decision maker to understand the differences between solutions and make an Informed decision. We show that in the bioinformatics domain of aligning protein/DNA sequences, a small set of near-optimal alignments are closer to the biologically known alignment than the optimal solution returned by the sequence alignment algorithms. Thus, we have developed a system for generating, viewing and comparing solutions among a set of near-optimal alternatives.