Stochastic Optimization of Engineering Drawing Analysis

by
Chimiao Xu

Advised by
Thomas Henderson

Automatic domain knowledge guided engineering drawing analysis systems have been widely exploited using computer science technology. My research focus on stochastic optimization of engineering drawing analysis process. The fundamental concept is applying the principles and theories of nondeterminism in multiple independent programs (agents) to explore threshold space and algorithm space which are tedious work for human. The system's ultimate goal is that the NonDeterministic Agent System (NDAS) should be able to self-guide its agents' actions along with proper threshold parameter values based on explicit knowledge of the engineering drawing and analysis process. Several stochastic optimization techniques, such as applying histgrams data on a set of features of standard engineering drawing, or other domain knowledge will be discussed.