AComparison Analysis of Conventional Classifiers and Deep Learning Models for Activity Recognition in Smart Homes

Activity recognition is essential for exploring human activities in smart homes in the presence of multiple sensors as residents interact with household appliances. smart homes use intelligent iot devices linked to residents' homes to track human behavior as humans interact with the home's equipment, which may improve healthcare and security issues for the residents. although remarkable studies have been done for pattern recognition and prediction of human activities in smart homes based on single residents and multiple residents using wearable sensors. however, not much research has been done on using activity recognizing ambient sensing (aras) residents. in this paper, we suggested using the aras dataset and newly emerged algorithms such as deep learning models to predict the activities of daily living (adl). we compared the performance of deep learning models (ann, cnn, and rnn) with that of classification models (dt, lda, adaboost, gb, xgboost, mpl, and knn) to figure out the adl in the smart home residents. the experimental results demonstrated that dl models outperformed with an excellent accuracy compared to conventional classifiers in houses a and b in recognizing adl in smart homes. this work proves that deep learning models perform best in analyzing aras datasets compared to traditional machine learning algorithms.