Neural network committee-based sensitivity analysis strategy for geotechnical engineering problems

摘要: Neural network usually acts as a “black box” in diverse fields to perform prediction, classification, and regression. Different from the conventional usages, neural is herein attempted handle factor sensitivity analysis geotechnical engineering system. After systematically investigating instability of employing single analysis, committee (NNC)-based strategy first algorithmically presented based on particular mathematical ideas weak law large numbers probability optimization. Significantly, this study especially emphasizes practical application NNC-based highlight mechanism underlying strata movement. The principal goal reveal relationships among influential factors movement through estimating relative contribution each explicative (input) variable dependent (output) variables It demonstrated that rationally not only reveals but also indicates predictability variable. In addition, an improved prediction model resulted integrating results into modeling, it capable facilitating convergence training advancing its precision angles. above outcomes indicate provides new paradigm applying networks deal with complex problems.