Open AccessMachine Learning Vs Deep Learning: Which Is Better For Human Activity Recognition
Author(s) -
Nasim Uddin,
Mohit Singh,
Mugalodi Rakesh
Publication year - 2020
Publication title -
international journal of engineering and advanced technology
Language(s) - English
Resource type - Journals
ISSN - 2249-8958
DOI - 10.35940/ijeat.c6310.049420
Subject(s) - confusion matrix , artificial intelligence , activity recognition , accelerometer , machine learning , confusion , computer science , deep learning , sitting , robotics , entertainment , gyroscope , human–computer interaction , engineering , robot , psychology , medicine , art , pathology , psychoanalysis , visual arts , aerospace engineering , operating system
Human activity recognition(HAR) is used to describe basic activities that humans are performing using the sensors that we have in smartphones. The data for this activity recognition is captured by various sensors of mobile phones or wristbands such as accelerometer, gyroscope and gravity sensors.HAR has grabbed the attention of various researchers due to its vast demand in the fields of sport training, security, entertainment health monitoring,computer vision and robotics. In this project we compare different machine learning and deep learning algorithms to find a better approach for HAR. The dataset comprises six activities i.e. walking, sleeping, sitting,moving upward, moving downwards and standing.In this demonstration we also showed confusion matrix,accuracy and multi log loss of various algorithms. With the help of accuracy, confusion matrix of algorithms we compare and determine the best approach for HAR. This will help in future research to map the activities of humans using one of the best approaches used