The twin issues of loss quality (accuracy) and training time are critical in choosing a stochastic optimizer for training deep neural networks. Optimization methods for machine learning include gradient descent, simulated annealing, genetic algorithm and second order techniques like Newton’s method. However, the popular method for optimizing neural networks is gradient descent. Overtime, researchers have made gradient descent more responsive to the requirements of improved quality loss (accuracy) and reduced training time by progressing from using simple learning rate to using adaptive moment estimation technique for parameter tuning. In this work, we investigate the performances of established stochastic gradient descent algorithms like Adam, RMSProp, Adagrad, and Adadelta in terms of training time and loss quality. We show practically, using series of stochastic experiments, that adaptive moment estimation has improved the gradient descent optimization method. Based on the empirical outcomes, we recommend further improvement of the method by using higher moments of gradient for parameter tuning (weight update). The output of our experiments also indicate that neural network is a stochastic algorithm.

In this work, we adopt an engineering problem-solving approach to the open-air defecation health problem. We model social and behaviour change communication intervention among other components of a water-sanitation-hygiene (WASH) system in response to the menace of open defecation in rural and urban communities globally. We also used experimental outcomes to show empirically that patterns in data captured in the WASH process could be learnt for effective decision making using deep learning neural networks as an intelligent software engineering technique. Eradicating open defecation is one of the indicators used for measuring progress made towards the attainment of Sustainable Development Goal 6 (SDG 6). We use the Adum-Aiona community in Nigeria as case study in designing community-based total sanitation programs using software model-driven engineering approaches with the aim of promoting their implementation. This is because even when toilets and other sanitary infrastructure are available, behavior and social change efforts are needed to promote their large-scale use. Also, we demonstrate that besides being used to model software systems, computational models (software architecture) are useful in documenting and promoting understanding of concepts in virtually all fields of human endeavour. Our motivation is that enhancing understanding of open defecation through software modelling would help SDG 6 implementors and actors attain set sanitation goals in both rural and urban communities towards the SDGs target year 2030.