Tiny Machine Learning

• Record Nr.: UNINA9911054615103321
• Autore: Imoize Agbotiname Lucky
• Pubbl/distr/stampa: Newark : , : John Wiley & Sons, Incorporated, , 2025; ©2026
• ISBN: 1-394-29455-7; 1-394-29456-5; 1-394-29457-3
• Edizione: [1st ed.]
• Descrizione fisica: 1 online resource (778 pages)
• Disciplina: 006.31
• Soggetti: Machine learning; Artificial intelligence
• Lingua di pubblicazione: Inglese
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Nota di contenuto: Cover -- Title Page -- Copyright -- Contents -- About the Editors -- List of Contributors -- Preface -- Chapter 1 Introduction to TinyML -- 1.1 Introduction -- 1.1.1 Key Contributions of the Chapter -- 1.1.2 Chapter Organization -- 1.2 Evolution of TinyML -- 1.3 Key Milestones and Current Trends -- 1.4 TinyML System Development -- 1.4.1 Hardware Platforms -- 1.4.2 Software Frameworks -- 1.5 Challenges and Bottlenecks -- 1.6 Cost-Benefit Analysis -- 1.7 Key Findings -- 1.8 Limitations of TinyML -- 1.9 Conclusion -- References -- Chapter 2 Learning Panorama Under TinyML -- 2.1 Introduction -- 2.1.1 Chapter Contribution -- 2.1.2 Chapter Organization -- 2.2 Challenges and Opportunities for Improved TinyML Model Design -- 2.2.1 Neural Architecture Search -- 2.2.2 Neural Attention Mechanisms -- 2.2.3 Depth‐Separable Convolution -- 2.2.4 Rational Activation Functions -- 2.3 Frontiers in Model Optimization for TinyML -- 2.3.1 Pruning -- 2.3.2 Quantization -- 2.3.3 Low‐Rank Decomposition -- 2.3.4 Knowledge Distillation -- 2.4 Learning Frameworks and Tools for TinyML Development -- 2.5 Frontiers for Algorithmic Innovations for TinyML -- 2.5.1 Multimodal Machine Learning for TinyML -- 2.5.2 Meta‐Learning Algorithms for TinyML -- 2.6 TinyML Development Process -- 2.7 Key Findings -- 2.8 Conclusion -- References -- ; Chapter 3 TinyML for Anomaly Detection -- 3.1 Introduction -- 3.1.1 Research Concepts and Motivation -- 3.1.2 Importance of TinyML in Anomaly Detection -- 3.1.3 Key Contributions -- 3.1.4 Chapter Organization -- 3.2 Context and Literature Review -- 3.2.1 TinyML: An Overview -- 3.2.2 Definition and Key Features -- 3.2.3 Advantages of TinyML for Edge Devices -- 3.2.4 Limitations and Challenges of TinyML -- 3.2.4.1 Hardware Constraints -- 3.2.4.2 Model Optimization Challenge -- 3.2.4.3 Data Availability and Preprocessing Constraints.; 3.2.4.4 Deployment Challenges -- 3.2.4.5 Development and Testing Challenges -- 3.2.5 Related Work -- 3.2.6 Anomaly Detection in IoT and Edge Devices -- 3.2.7 Structured Literature Mapping Approach -- 3.2.7.1 Understanding PICOC in the Context of Systematic Literature Mapping -- 3.2.7.2 Formulating Research Questions Based on the PICOC Framework -- 3.2.7.3 Inclusion/Exclusion Criteria -- 3.2.7.4 Search Strategy -- 3.2.7.5 Data Extraction Process -- 3.2.7.6 Distribution of Research Studies -- 3.2.7.7 Case Study 1: Industrial Anomaly Detection Using Edge Computing and TinyML -- 3.2.7.8 Case Study 2: TinyML for Anomaly Detection in Industrial Equipment Monitoring -- 3.2.7.9 Case Study 3: TinyML for Anomaly Detection in Smart Agriculture -- 3.2.7.10 Case Study 4: TinyML for Anomaly Detection in Smart Homes -- 3.2.7.11 Case Study 4: TinyML for Anomaly Detection in Wearable Health Devices -- 3.2.8 Machine Learning Models for Anomaly Detection -- 3.2.9 TinyML Applications in Anomaly Detection -- 3.2.10 Practical Applications of TinyML in Anomaly Detection -- 3.2.10.1 Industrial Machinery Monitoring -- 3.2.10.2 Infrastructure Health Monitoring -- 3.2.10.3 Cybersecurity in IoT Networks -- 3.2.10.4 Environmental Monitoring -- 3.2.10.5 Healthcare and Wearable Devices -- 3.2.10.6 Smart Cities and Transportation Systems -- 3.3 Lessons Learned -- 3.3.1 Adaptability of Models to Constrained Environments -- 3.3.2 Importance of Real‐Time Processing -- 3.3.3 Balancing Precision and Efficiency -- 3.3.4 Data Quality and Preprocessing -- 3.3.5 Privacy and Security Concerns -- 3.3.6 Scalability and Deployment Challenges -- 3.3.7 Continuous Learning and Adaptation -- 3.3.8 Collaborative and Interdisciplinary Approaches -- 3.3.9 Transformative Potential of TinyML -- 3.3.10 Open Challenges -- 3.4 Future Scope -- 3.4.1 Model Optimization and Transfer Learning.; 3.4.2 Energy‐Efficient Algorithms -- 3.4.3 Federated and Collaborative Learning -- 3.4.4 Real‐Time and Adaptive Detection -- 3.4.5 Multimodal Data Fusion -- 3.4.6 Security and Robustness -- 3.4.7 Integration with Emerging Technologies -- 3.4.8 Applications in New Domains -- 3.5 Conclusion -- References -- Chapter 4 TinyML Power Consumption and Memory in IoT MCUs -- 4.1 Introduction -- 4.1.1 Background and Context of IoT MCUs -- 4.1.2 Emergence of TinyML in IoT Systems -- 4.1.3 Power Consumption Challenges in IoT MCUs -- 4.1.4 Memory Constraint in IoT MCUs -- 4.1.5 Ultralow‐Power Smart IoT Devices with Embedded TinyML -- 4.1.6 Data Management Techniques Utilizing TinyML in IoT Systems -- 4.1.7 TinyML for Real‐Time Low‐Power IoT Circuit Modeling -- 4.1.8 TinyML for Ultralow‐Power AI in Large‐Scale IoT Deployments -- 4.1.9 Research Contributions -- 4.2 Context and Literature Review -- 4.2.1 Power Consumption in IoT MCUs -- 4.2.2 Memory Constraints in IoT MCUs -- 4.2.3 Integration of Power and Memory Optimization Techniques -- 4.2.4 Comparative Analysis of MCU Architectures for IoT -- 4.3 Methodology -- 4.3.1 Research Design: Literature Review Framework -- 4.3.2 Defining the Scope of the Literature Review -- 4.3.3 Developing Selection Criteria -- 4.3.3.1 Inclusion and Exclusion Criteria -- 4.3.4 Search Strategy Implementation -- 4.3.5 Data Extraction and ; Organization -- 4.3.6 Synthesizing Findings -- 4.3.7 Review Process and Analysis -- 4.4 Results and Discussion: Bibliometric Analysis -- 4.4.1 Publication Trends -- 4.4.2 Collaborative Networks -- 4.4.3 Thematic Focus Areas -- 4.4.4 Insights from Visualization -- 4.4.5 Discussions -- 4.5 Conclusion and Future Directions -- References -- Chapter 5 Efficient Data Cleaning and Anomaly Detection in IoT Devices Using TinyCleanEDF -- 5.1 Introduction -- 5.2 IoT and TinyCleanEDF.; 5.3 The Importance of Data Cleaning in IoT Systems -- 5.4 Anomaly Detection and Its Importance in IoT Systems -- 5.4.1 Anomaly Detection -- 5.5 Data Preprocessing and Cleaning Workflow -- 5.5.1 Noise Removal Techniques -- 5.5.2 Handling Missing Data -- 5.5.3 Anomaly Detection Framework -- 5.5.3.1 Statistical Approaches -- 5.5.3.2 Machine Learning Algorithms -- 5.5.4 Federated Learning Integration -- 5.5.5 Distributed Model Training -- 5.5.6 Data Privacy Mechanisms -- 5.5.7 Scalability and Compatibility Enhancements -- 5.6 Implementation Details -- 5.6.1 Software and Hardware Requirements -- 5.6.1.1 Software Requirements -- 5.6.1.2 Hardware Requirements -- 5.6.2 System Architecture of TinyCleanEDF -- 5.6.2.1 Data Acquisition Layer -- 5.6.2.2 Local Processing Layer -- 5.6.2.3 Federated Aggregation Layer -- 5.6.2.4 Communication Layer -- 5.6.3 Deployment on IoT Devices -- 5.6.3.1 Pre‐deployment Preparation -- 5.6.3.2 Installation and Initialization -- 5.6.3.3 Operational Phase -- 5.7 Case Studies and Applications -- 5.8 Future Directions -- 5.8.1 Enhancements in Federated Learning Algorithms -- 5.8.2 Integration with Edge AI Technologies -- 5.8.3 Expanding to Other IoT Use Cases -- 5.9 Conclusion and Future Scope -- References -- Chapter 6 TinyML Devices and Tools -- 6.1 Introduction -- 6.1.1 Key Contributions of the Chapter -- 6.1.2 Chapter Organization -- 6.2 Related Work -- 6.3 TinyML Devices -- 6.3.1 Architecture and Functionality -- 6.4 TinyML Tools -- 6.4.1 TensorFlow Lite for Microcontrollers (TFLM) -- 6.4.2 Edge Impulse -- 6.4.3 PyTorch Mobile -- 6.4.4 TensorFlow Lite (TFL) -- 6.4.5 uTensor -- 6.4.6 CMSIS‐NN -- 6.4.7 STM32Cube.AI and NanoEdge Studio -- 6.4.8 EmbML -- 6.4.9 FANN on MCU -- 6.5 Deployment Procedure of TinyML -- 6.5.1 Case Studies of TinyML in the Health Sector Domain -- 6.5.2 Case Studies of TinyML in the Agriculture Domain.; 6.5.2.1 Irrigation Controller -- 6.6 Lesson Learned and Prospects -- 6.6.1 Lesson Learned -- 6.6.2 Prospects -- 6.7 Conclusion -- References -- Chapter 7 Privacy‐Preserving Techniques in TinyML for IoT -- 7.1 Introduction -- 7.2 Related Works -- 7.3 Homomorphic Encryption in TinyML -- 7.3.1 Fundamentals of Homomorphic Encryption -- 7.3.2 Partially and Fully Homomorphic Encryption Schemes -- 7.3.3 Applying Homomorphic Encryption to TinyML Tasks -- 7.3.4 Performance Evaluation and Practical Considerations -- 7.3.5 Comparison of Homomorphic Encryption Schemes -- 7.4 Differential Privacy for TinyML -- 7.4.1 Principles of Differential Privacy -- 7.4.2 Mechanisms for Achieving Differential Privacy -- 7.4.3 Differentially Private TinyML Algorithms -- 7.4.4 Balancing Privacy and Utility in TinyML Applications -- 7.5 Secure Multi‐Party Computation in TinyML -- 7.5.1 Foundations of Secure Multi‐Party Computation -- 7.5.2 Secret Sharing and Garbled Circuits -- 7.5.3 Secure Aggregation for Distributed TinyML -- 7.5.4 Efficiency Optimization for Secure Multi‐Party Computation in TinyML -- 7.6 Case Studies and Applications of Privacy‐Preserving TinyML -- 7.6.1 Efficient Secure Multi‐Party Computation for Privacy‐Preserving Machine Learning -- 7.6.2 A Generic Framework for Privacy‐Preserving Deep Learning -- 7.6.3 Maliciously Secure Coopetitive Learning for Linear Models -- 7.6.4 Practical Secure Aggregation for Privacy‐Preserving Machine Learning -- ; 7.6.5 Falcon: Honest Majority Maliciously Secure Framework for Private Deep Learning -- 7.7 Discussion and Future Directions -- 7.7.1 Secure Multi‐party Computation for TinyML -- 7.7.2 Homomorphic Encryption in TinyML -- 7.7.3 Differential Privacy for TinyML -- 7.7.4 Practical Considerations and Future Directions -- 7.8 Conclusion -- Acknowledgment -- References.; Chapter 8 Enhancing Cybersecurity in TinyML with Lightweight Cryptographic Algorithms.
• Sommario/riassunto: An expert compilation of on-device training techniques, regulatory frameworks, and ethical considerations of TinyML design and development In Tiny Machine Learning: Design Principles and Applications, a team of distinguished researchers delivers a comprehensive discussion of the critical concepts, design principles, applications, and relevant.