Human Activity Recognition with Smartphone Sensors Using RNN

In recent time, different hardware, smart devices, machine learning model and deep learning model are used for Human Activity Recognition (HAR) applications. The main implication of this question is that which creates HAR models with high accuracy, uses low cost. Smartphones have become an important part of persons life; they play vital role in human activity recognition systems to detect human actions. To recognize human activity, this article proposes the use of Logistic Regression and Recurrent Neural Networks (RNNs), which is used for handling time series data or sequential data. The long-term memory network model recognizes six basic human activities: sitting, standing, walking up and down, lying down, and walking. The LSTM model is an advanced form of Recurrent Neural Network that eliminates leaky gradients and burst gradients of RNN. It helps us to examine past information to perform current tasks or give output based on current input.