The Information processing based on data mining in WSN is at its starting stage, when compared to traditional machine learning in WSN. In order to solve a particular problem in WSN the researchers now-a-day were mainly focused on applying machine learning techniques. The different researchers will have different assumptions, application scenarios and preferences in applying machine learning algorithms. These differences will result to a major challenge in allowing researchers to build upon each other's work so that research results will accumulate in the community. Thus, a common architecture across the WSN machine learning community would be necessary in order to overcome these differences.