Activity Recognition and Prediction for Smart IoT Environments (Internet of Things)

✍ Scribed by Michele Ianni (editor), Antonella Guzzo (editor), Raffaele Gravina (editor), Hassan Ghasemzadeh (editor), Zhelong Wang (editor)

• Publisher: Springer
• Year: 2024
• Tongue: English
• Leaves: 188
• Edition: 2024
• Category: Library

✦ Synopsis

This book provides the latest developments in activity recognition and prediction, with particular focus on the Internet of Things. The book covers advanced research and state of the art of activity prediction and its practical application in different IoT related contexts, ranging from industrial to scientific, from business to daily living, from education to government and so on. New algorithms, architectures, and methodologies are proposed, as well as solutions to existing challenges with a focus on security, privacy, and safety. The book is relevant to researchers, academics, professionals and students.

✦ Table of Contents

Preface
Contents
Discovering Human Habits Through Process Mining: State of the Art and Research Challenges
1 Introduction
2 Background and Related Works
2.1 Sensor Data in Smart Spaces
2.2 Process Mining
2.3 Unsupervised Approaches to Ambient Intelligence
3 Unsupervised Human Habit Discovery
4 Experiments
5 Conclusion
Appendix
References
Methodology for Human Activity Recognition Based on Wearable Sensor Networks
1 Introduction
2 System Overview
2.1 System Overview
2.2 Definition and Conversion Relationship of Coordinate System
2.3 Algorithm Design and Validation of Motion Capture System
3 Data Processing and Model Evaluation
3.1 Data Processing and Model Evaluation
3.2 Feature Extraction
3.3 Feature Selection
3.4 Activity Recognition Algorithm
3.5 Experimental Results and Discussion
4 Applications of Wearable Inertial Sensor Networks
4.1 Human–Human Interactional Synchrony Analysis
4.2 Smart Healthcare-Wearable Gait Analysis for Parkinson's Disease
4.3 Intelligent Sports Performance Analysis: Investigating Horse-Rider Interaction through Body Sensor Network
5 Summary
References
A Sitting Posture Monitoring System in Wheelchair Users
1 Introduction
2 Related Works
2.1 Postural Monitoring Devices
2.2 Machine Learning Techniques Used for Posture Recognition
3 i-KuXin: New Sitting Postural Monitoring System
3.1 Monitoring Device Design
3.2 Acquisition System
4 Methodology of Experimental Trial Design
4.1 Definition of Experimental Tests
4.2 Database Generation
4.3 Data Preprocessing
5 Design of the Intelligent Postural Recognition System
5.1 Postural Recognition Intelligent Techniques Selection
5.2 System Training
6 Results and Discussion
6.1 Comparison of Results According to Technique
6.2 Analysis of the Optimal Number of Sensors
7 Conclusion
References
A Comprehensive Review of Deep Learning for Activity Recognition
1 Introduction
1.1 Challenges in HAR
1.2 Deep Learning in HAR
1.3 Contributions
2 Datasets
2.1 Sensory Data
Wearable Sensors
Ambient Sensors
Object Sensors
Hybrid Sensors
3 Challenges and Role of Deep Learning in HAR
3.1 Feature Extraction
Temporal Features
Multimodal Features
3.2 Annotation Scarcity
Unsupervised Learning
Semi-Supervised Learning
3.3 Class Imbalance
3.4 Distribution Discrepancies
3.5 Composite Activity
Fused Model
Hierarchical Model
3.6 Concurrent Activity
3.7 Multi-Occupant Activity
3.8 Computational Cost
3.9 Privacy
3.10 Explainability and Interpretability
4 Conclusion
References
Multi-User Activity Monitoring Based on Contactless Sensing
1 Introduction
2 Related Work
2.1 Bodily Sensing Based on Wi-Fi
2.2 Multi-User Activity Recognition with Wi-Fi Signals
3 Proposed Method
3.1 Mathematical Modeling of the Human Activity based on CSI Data
3.2 Processing Workflow
4 Experiments and Results
4.1 Experiment Setup
4.2 Experimental Results and Discussion
5 Conclusion
References
Efficient Sensing and Classification for Extended Battery Life
1 Introduction
2 Problem Statement
3 Method
3.1 Sensing and Computation Efficiency
Sampling Frequency Determination
Feature Selection
3.2 Cascading Classifier
Binary Classifiers
Groups Classifier
Classifier Flow
3.3 Resource Analysis
4 Validation
4.1 Experimental Setup
4.2 Binary Group Intensity classifiers
4.3 Multi-Class Classification of Activities
4.4 Cascading Classifier Performance
4.5 PAMP2 Dataset Evaluation
4.6 Power Consumption and Memory Usage Evaluation
4.7 Comparison with State-of-the Art
5 Conclusion, Discussion, and Future Work
References
Unveiling the Potential of Machine Learning in Activity Recognition for Industry 4.0
1 Introduction
2 Background
2.1 Machine Learning
2.2 Human Activity Recognition
3 Literature Review
3.1 Overview of ML Techniques for Activity Recognition
3.2 Benefits and Challenges of Using ML in Activity Recognition
4 Impact and Applications of HAR in Industry 4.0
4.1 Integration of AR and Industry 4.0
4.2 Case Studies
5 Future Directions and Conclusions
5.1 Challenges and Future Directions
6 Conclusions
References
Human Activity Recognition: Trends and Challenges
1 Introduction
1.1 Contributions
2 Trends and Challenges Related to Data in HAR
2.1 Sensor-Based HAR
2.2 Vision-Based HAR
3 Trends and Challenges in HAR System Implementation
3.1 Data Preprocessing
3.2 Segmentation
3.3 Feature Engineering
3.4 Model Selection for Training
4 Conclusion
References
Index