Hands-On Artificial Intelligence for IoT - Second Edition

Just like the OSI reference model for the internet, IoT architecture is defined through six layers: four horizontal layers and two vertical layers. The two vertical layers are Management and Security and they're spread over all four horizontal layers, as seen in the following diagram:

IoT layers

The Device Layer: At the bottom of the stack, we have the device layer, also called the perception layer. This layer contains the physical things needed to sense or control the physical world and acquire data (that is, by perceiving the physical world). Existing hardware, such as sensors, RFID, and actuators, constitutes the perception layer. 

The Network Layer: This layer provides the networking support and transfer of data over either wired or wireless network. The layer securely transmits the information from the devices in the device layer to the information processing system. Both transmission MediumandTechnology are part of the networking layer. Examples include 3G, UMTS, ZigBee, Bluetooth, Wi-Fi, and so on.

The Service Layer: This layer is responsible for service management. It receives information from the network layer, stores it into the database, processes that information, and can make an automatic decision based on the results. 

The Application Layer: This layer manages the applications dependent upon the information processed in the service layer. There's a wide range of applications that can be implemented by IoT: smart cities, smart farming, and smart homes, to name a few. 

Information from the network layer is often managed with the help of IoT platforms. Many companies today provide IoT platform services, where they help not only with data but also enable seamless integration with different hardware. Since they function as a mediator between the hardware and application layer, IoT platforms are also referred to as IoT middleware and are part of the service layer in the IoT reference stack. IoT platforms provide the ability to connect and communicate with things from anywhere in the world. In this book, we'll briefly cover some popular IoT platforms such as the Google Cloud Platform, Azure IoT, Amazon AWS IoT, Predix, and H2O. 

You can select which IoT platform is best for you based on the following criteria:

A vertical market is a market in which vendors offer goods and services specific to an industry, trade, profession, or other groups of customers with specialized needs. IoT enables the possibility of many such verticals, and some of the top IoT verticals are as follows:

We'll be covering some of them as a case study in this book. The content of this book is focused on information processing and the applications being implemented on IoT and so we'll not be going into details of the devices, architecture, and protocols involved in IoT reference stacks any further.

For IoT problems, the most used data management (DM) methodology is cross-industry standard process for data mining (CRISP-DM) proposed by Chapman et al. It's a process model that states the tasks that need to be carried out for successfully completing DM. It's a vendor-independent methodology divided into these six different phases:

Following diagram shows the different stages:

 Different stages in CRISP-DM

As we can see, it's a continuous process model with data science and AI playing important roles in steps 2–5.

A large number of cloud platforms with both AI and IoT capabilities are available today. These platforms provide the capability to integrate the sensors and devices and perform analytics on the cloud. There exist more than 30 cloud platforms in the global market, each targeting different IoT verticals and services. The following screenshot lists the various services that AI/IoT platforms support:

 Services that different AI/IoT platforms support

Let's briefly find out about some popular cloud platforms. In Chapter 12, Combining it all together, we'll learn how to use the most popular ones. The following is a list of some of the popular Cloud platforms:

TensorFlow is an open source software library developed by the Google Brain team; it has functions and APIs for implementing deep neural networks. It works with Python, C++, Java, R, and Go. It can be used to work on multiple platforms, CPU, GPU, mobile, and even distributed. TensorFlow allows for model deployment and ease of use in production. The optimizer in TensorFlow makes the task of training deep neural networks easier by automatically calculating gradients and applying them to update weights and biases.

In TensorFlow, a program has two distinct components:

Let's see the code to perform matrix multiplication in TensorFlow. The whole code can be accessed from the GitHub repository (https://github.com/PacktPublishing/Hands-On-Artificial-Intelligence-for-IoT) filename, matrix_multiplication.ipynb: 

import tensorflow as tf
import numpy as np

This part imports the TensorFlow module. Next, we define the computation graph. mat1 and mat2 are two matrices we need to multiply:

# A random matrix of size [3,5]
mat1 = np.random.rand(3,5)  
# A random matrix of size [5,2]
mat2 = np.random.rand(5,2)  

We declare two placeholders, A and B, so that we can pass their values at runtime. In the computation graph, we declare all of the data and computation objects:

# Declare placeholders for the two matrices 
A = tf.placeholder(tf.float32, None, name='A')
B = tf.placeholder(tf.float32, None, name='B')  

This declares two placeholders with the names A and B; the arguments to the tf.placeholder method specify that the placeholders are of the float32 datatype. Since the shape specified is None, we can feed it a tensor of any shape and an optional name for the operation. Next, we define the operation to be performed using the matrix multiplication method, tf.matmul:

C = tf.matmul(A,B)

The execution graph is declared as a Session object, which is fed the two matrices, mat1 and mat2, for the placeholders, A and B, respectively:

with tf.Session() as sess:
    result = sess.run(C, feed_dict={A: mat1, B:mat2})
    print(result)

Keras is a high-level API that runs on top of TensorFlow. It allows for fast and easy prototyping. It supports both convolutional and recurrent neural networks, and even a combination of the two. It can run on both CPUs and GPUs. The following code performs matrix multiplication using Keras:

# Import the libraries
import keras.backend as K
import numpy as np

# Declare the data
A = np.random.rand(20,500)
B = np.random.rand(500,3000)

#Create Variable
x = K.variable(value=A)
y = K.variable(value=B)
z = K.dot(x,y)
print(K.eval(z))

In the coming chapters, we'll be learning different DL models and ML methods. They all work on data; while a large number of datasets are available to demonstrate how these models work, in this book, we'll use datasets available freely through wireless sensors and other IoT devices. The following are some of the datasets used in this book and their sources.