This book provides an in-depth analysis of the design and implementation of a smartphone sensor-based Human Activity Recognition (HAR) system using advanced learning models. It offers detailed insights into key advancements and future directions for HAR system development. The content includes a comprehensive literature review on wearable sensor-based HAR, along with descriptions of publicly available datasets for building HAR systems. To overcome the limitations of existing sensor-based HAR systems, the book emphasizes the creation of a robust HAR dataset in uncontrolled environments. This dataset is developed using built-in smartphone sensors, enabling real-time activity classification. The book presents multiple effective and optimized classifiers for developing efficient HAR systems, leveraging both self-generated and publicly available data.
The book demonstrates high effectiveness in real-time activity classification and addresses common challenges such as class imbalance and domain shift. It provides practical methodologies and techniques to enhance HAR system performance. This book is an essential resource for professionals, healthcare practitioners, academics, researchers, and students specializing in health applications, signal processing, machine learning, ensemble learning, and deep learning. It offers extensive analysis and actionable insights for investigators, helping streamline product development through intelligent learning algorithms in the field of Human Activity Recognition.
Table of contents (8 chapters)
Front Matter
Pages i-xv
Introduction
Pages 1-8
Literature Survey
Pages 9-23
Design and Development of a Comprehensive Smartphone Sensor-Based HAR Dataset in an Uncontrolled Environment for Efficient HAR
Pages 25-35
Data Intensity-Based Feature Selection Protocol for Sensor-Based HAR System
Pages 37-56
Comparative Analysis of Shallow, Ensemble, and Deep Learning Models for Sensor-Based HAR in Uncontrolled Environments
Nurul Amin Choudhury, Badal Soni
Pages 57-68
Optimized Deep Learning for HAR in Uncontrolled Environments with Hybrid Feature Fusion
Pages 69-81
A Variable Batch Size-Based Hybrid Deep Learning Framework for HAR in Uncontrolled Environments
Pages 83-98
Conclusion and Future Works
Nurul Amin Choudhury, Badal Soni
Pages 99-101
