摘要：Marine container transport is now an indispensable mode of transport for the global economy. For Japan, the import and export of goods by marine containers is extremely important for social and economic activities, too. To improve the transportation effectively it is very useful to develop a methods which enables detailed and quantitative analysis of the impact of time, cost, and other diverse transport attributes on the choice of transport route. On the other hand, research on artificial intelligence (AI) has been progressing rapidly in recent years. There are some fields in which cognition ability by AI exceeds that by human. In this research, we take up the problem of estimating the transportation route of imported and exported marine container cargo which originates and arrives in Japan. We try to develop the route selection method using the Deep Learning which is one of the main technologies in AI research field. This approach takes into account nonlinear and complex relationships among various transport attributes. It is expected that detailed examination of route performance will become possible based on the appropriateness of route for individual cargo.
