Book Image

Hands-On Deep Learning for IoT

By : Dr. Mohammad Abdur Razzaque, Md. Rezaul Karim
Book Image

Hands-On Deep Learning for IoT

By: Dr. Mohammad Abdur Razzaque, Md. Rezaul Karim

Overview of this book

Artificial Intelligence is growing quickly, which is driven by advancements in neural networks(NN) and deep learning (DL). With an increase in investments in smart cities, smart healthcare, and industrial Internet of Things (IoT), commercialization of IoT will soon be at peak in which massive amounts of data generated by IoT devices need to be processed at scale. Hands-On Deep Learning for IoT will provide deeper insights into IoT data, which will start by introducing how DL fits into the context of making IoT applications smarter. It then covers how to build deep architectures using TensorFlow, Keras, and Chainer for IoT. You’ll learn how to train convolutional neural networks(CNN) to develop applications for image-based road faults detection and smart garbage separation, followed by implementing voice-initiated smart light control and home access mechanisms powered by recurrent neural networks(RNN). You’ll master IoT applications for indoor localization, predictive maintenance, and locating equipment in a large hospital using autoencoders, DeepFi, and LSTM networks. Furthermore, you’ll learn IoT application development for healthcare with IoT security enhanced. By the end of this book, you will have sufficient knowledge need to use deep learning efficiently to power your IoT-based applications for smarter decision making.

Related Content you might be interested in

Current Title:

Small book image
Hands-On Deep Learning for IoT
Dr. Mohammad Abdur Razzaque | Md. Rezaul Karim
Jun, 2019 | 308 pages
Small book image
Machine Learning with Scala Quick Start Guide
Md Rezaul Karim
Apr, 2019 | 220 pages
Small book image
Java Deep Learning Projects
Md Rezaul Karim
Jun, 2018 | 436 pages
Small book image
Deep Learning with TensorFlow
Md Rezaul Karim | Giancarlo Zaccone
Mar, 2018 | 484 pages
Table of Contents (15 chapters)
Preface
Preface
Who this book is for
What this book covers
To get the most out of this book
Get in touch
Free Chapter
1
Section 1: IoT Ecosystems, Deep Learning Techniques, and Frameworks
Section 1: IoT Ecosystems, Deep Learning Techniques, and Frameworks
2
The End-to-End Life Cycle of the IoT
The End-to-End Life Cycle of the IoT
The E2E life cycle of the IoT
IoT application domains
The importance of analytics in IoT
The motivation to use DL in IoT data analytics
The key characteristics and requirements of IoT data
Summary
Reference
3
Deep Learning Architectures for IoT
Deep Learning Architectures for IoT
A soft introduction to ML
ML tasks
Delving into DL
Artificial neural networks
Neural network architectures
DL frameworks and cloud platforms for IoT
Summary
4
Section 2: Hands-On Deep Learning Application Development for IoT
Section 2: Hands-On Deep Learning Application Development for IoT
5
Image Recognition in IoT
Image Recognition in IoT
IoT applications and image recognition
Use case one – image-based automated fault detection
Use case two – image-based smart solid waste separation
Transfer learning for image recognition in IoT
CNNs for image recognition in IoT applications
Collecting data for use case one
Collecting data for use case two
Data pre-processing
Models training
Evaluating models
Summary
References
6
Audio/Speech/Voice Recognition in IoT
Audio/Speech/Voice Recognition in IoT
Speech/voice recognition for IoT
Use case one – voice-controlled smart light
Use case two – voice-controlled home access
DL for sound/audio recognition in IoT
CNNs and transfer learning for speech recognition in IoT applications
Collecting data
Data preprocessing
Models training
Evaluating models
Summary
References
7
Indoor Localization in IoT
Indoor Localization in IoT
An overview of indoor localization
DL-based indoor localization for IoT
Example – Indoor localization with Wi-Fi fingerprinting
Deployment techniques
Summary
8
Physiological and Psychological State Detection in IoT
Physiological and Psychological State Detection in IoT
IoT-based human physiological and psychological state detection
Use case one – remote progress monitoring of physiotherapy
Use case two — IoT-based smart classroom
Deep learning for human activity and emotion detection in IoT
LSTM, CNNs, and transfer learning for HAR/FER in IoT applications
Data collection
Data exploration
Data preprocessing
Model training
Model evaluation
Summary
References
9
IoT Security
IoT Security
Security attacks in IoT and detections
Use case one: intelligent host intrusion detection in IoT
Use case two: traffic-based intelligent network intrusion detection in IoT
DL for IoT security incident detection
Data collection
Data preprocessing
Model training
Model evaluation
Summary
References
10
Section 3: Advanced Aspects and Analytics in IoT
Section 3: Advanced Aspects and Analytics in IoT
11
Predictive Maintenance for IoT
Predictive Maintenance for IoT
Predictive maintenance for IoT
Example – PM for an aircraft gas turbine engine
DL for predicting RLU
FAQs
Summary
12
Deep Learning in Healthcare IoT
Deep Learning in Healthcare IoT
IoT in healthcare
Use case one – remote management of chronic disease
Use case two – IoT for acne detection and care
DL for healthcare
Data collection
Data exploration
Data preprocessing
Model training
Model evaluations
Summary
References
13
What's Next - Wrapping Up and Future Directions
What's Next - Wrapping Up and Future Directions
What we have covered in this book?
Deployment challenges of DL solutions in resource-constrained IoT devices
Existing solutions to support DL in resource-constrained IoT devices
Potential future solutions
Summary
References
14
Other Books You May Enjoy
Other Books You May Enjoy
Leave a review - let other readers know what you think

Model evaluation

We can evaluate three different aspects of the models:

  • Learning/(re)training time
  • Storage requirement
  • Performance (accuracy)

On a desktop (Intel Xenon CPU E5-1650 [email protected] and 32 GB RAM) with GPU support, the training of LSTM on the CPU-utilization dataset and the autoencoder on the KDD layered wise dataset (reduced dataset) took a few minutes. The DNN model on the overall dataset took a little over an hour, which was expected as it has been trained on a larger dataset (KDD's overall 10% dataset).

The storage requirement of a model is an essential consideration in resource-constrained IoT devices. The following screenshot presents the storage requirements for the three models we tested for the two use cases:

As shown in the screenshot, autoencoders took storage in the range of KB. The final version of a stored autoencoder model took only 85 KB, LSTM...

Previous Section
End of Section 9
Next Section