The purpose of this study is to utilize Deep Learning technology which is actively developed and utilized recently in the process of building alternative design. This means that the architect examines situations that are difficult to think about and automatically presents the results to help architects make the best alternative decisions. This technology is different from the artificial intelligence, which is a simple method that pre-sets judgment criteria in the same way as the theme that has been studied in the field of traditional architectural design computing such as design automation, There is a difference in deriving the optimum result by reasoning. In this study, we aim to develop a technique that learns human behavior through space and object arrangement in plane design and derives optimal result by reasoning it. That is to say, the development of automatic alternative generation technology using deep learning technology for furniture and space arrangement of daycare center. In this study, we present an example of system development that automatically creates alternatives and suggests optimal alternatives by modeling the area of children s space and activities of children and introducing existing Deep Learning technology.