Deep neural network architectures for social services diagnosis in smart cities

Abstract

Abstract Social services intend to aid disadvantaged, distressed, or vulnerable persons or groups. Machine Learning (ML) and Deep Learning (DL), which are important technologies to leverage Internet of Things and Big Data, have not been considered to support intelligent social services in Smart Cities. Using technology to achieve more responsive, efficient, and proactive social services is a must in Smart Cities because it will lead to a more fair and egalitarian society. This research work contributes with the evaluation of a thousand Neural Networks architectures for the automatic diagnosis of chronic social exclusion. Some of them outperform previous models in quality metrics such as accuracy and F-score. Beyond the improvement in predicting this specific social condition, to the best of the authors’ knowledge, this paper open the research line of applying these methods for the general social services diagnosis in Smart Cities. Finally, the advantages of using the DL paradigm over other ML alternatives in this scope are discussed.