Ontology Based Activity Learning Using ANN For Producing Behavioral Models For Individual Inhabitants

Detection,recognition,and learning of daily life activities are one of the key components in the development of smart home technology.In this context,ontology-based activity learning and recognition model plays an important role im diverse domains i.e.smart homes,old homes,and hospitals.We encounter a few challenges with ontological models.Actually,those models are static in nature and can’t be self-evolution.This type of model can’t be completed at once and we can’t restrict the inhabitants of a smart home(SH)for limited activities.Inhabitants are unpredictable in nature and can perform those activities of daily life(ADL)that aren’t mentioned in the ontological model.Based on an adaptive Artificial Neural Network called ANN,this research presents a new computational approach.It uses a knowledge-based approach based on determining the real-time activity of multiple sensor data streams in a smart home.This is an ontology-based method,so in the context of activity modeling and expression representation,it goes beyond the traditional method of activity recognition.The ontology-based proposed method uses knowledge-based approach that has been implemented by data annotation and evaluated with in-depth experiments of enormous cases of activities of daily life.It is tested on a self-prepared data set,based on knowledge by determining the real-time activity of multiple sensor data streams in a smart home for approximately two months.The learning system uses the approach of Multi-Layer Perceptron(MLP)and annotated sensor’s dataset.The results show that MLP has some attractive features to analyze.There was a need to develop an interactive and easy-to-use data extraction system that analyzes human activities in a smart home environment through better pattern recognition,visualization and interpretation to produced behavioral models according to inhabitant activities of daily life.In short,the proposed system will learn from existing and new ADLs,that can’t be identified by the recognition model.In order to learn the newest inhabitant behavior to perform and record those new activities,we use ANN and adding new properties with existing activities.After compact process of ADLs recognition and learning,the purpose of producing behavioral model from generic model has been achieved in results and model was supposed to be learned and successfully capture the ADLs of individuals in form of actions with all behaviors.