KidNet

Kidney cancer is one of the 10 most common cancers in both men and women. The lifetime risk for one developing kidney cancer is about 1.6%. The rate of kidney cancer diagnosis has been rising since the 1990s due to the use of newer imaging tests such as CT scans. The kidneys are deep inside the body and hence small kidney tumours cannot be seen or felt during a physical examination. Existing work on kidney tumour diagnosis uses traditional machine learning and image processing techniques to find and classify the images. Deep learning systems do not require this domain-specific knowledge. The kidney tumour diagnosis system uses deep learning and convolutional neural networks to classify CT images. A deep learning neural network model named KidNet has been implemented. It has been trained using labelled kidney CT images. To achieve acceleration during the training phase, GPUs have been used. The network when trained with abdominal CT images achieved 86.1% accuracy, and the one trained with cropped portion of kidney images achieved 89.6% accuracy.

Facial Emotion Recognition System Using Entire Feature Vectors and Supervised Classifier

This chapter proposes the facial expression system with the entire facial feature of geometric deformable model and classifier in order to analyze the set of prototype expressions from frontal macro facial expression. In the training phase, the face detection and tracking are carried out by constrained local model (CLM) on a standardized database. Using the CLM grid node, the entire feature vector displacement is obtained by facial feature extraction, which has 66 feature points. The feature vector displacement is computed in bi-linear support vector machines (SVMs) classifier to evaluate the facial and develops the trained model. Similarly, the testing phase is carried out and the outcome is equated with the trained model for human emotion identifications. Two normalization techniques and hold-out validations are computed in both phases. Through this model, the overall validation performance is higher than existing models.

Deep Learning in IoT

The internet of things (IoT), big data analytics, and deep learning (DL) applications in the mechanical internet are expanding. The current digital era has various sensory devices for a wide range of fields and applications, which all generate various sensory data. DL is being applied for handling big data and has achieved great success in the IoT and other fields. The applications for data streams to discover new information, predict future insights, and make control decisions are crucial processes that make the IoT a worthy paradigm for businesses and a quality-of-life improving technology. This chapter provides a detailed account of the IoT domain, machine learning, and DL techniques and applications. The IoT that consists of DL with intelligence backgrounds is also discussed. Recent research on DL in the IoT within the big data domain is also discussed. Current challenges and potential areas for future research are discussed.

Deep Learning Solutions for Agricultural and Farming Activities

The continuously growing population throughout globe demands an ample food supply, which is one of foremost challenge of smart agriculture. Timely and precise identification of weeds, insects, and diseases in plants are necessary for increased crop yield to satisfy demand for sufficient food supply. With fewer experts in this field, there is a need to develop an automated system for predicting yield, detection of weeds, insects, and diseases in plants. In addition to plants, livestock such as cattle, pigs, and chickens also contribute as major food. Hence, livestock demands precision methods for reducing the mortality rate of livestock by identifying diseases in livestock. Deep learning is one of the upcoming technologies that when combined with image processing promises smart agriculture to be a reality. Various applications of DL for smart agriculture are covered.

Deep Neural Network-Based Android Malware Detection (D-AMD)

Android is an operating system that presently has over one billion active users for their mobile devices in which a copious quantity of information is available. Mobile malware causes security incidents like monetary damages, stealing of personal information, etc., when it's deep-rooted into the target devices. Since static and dynamic analysis of Android applications to detect the presence of malware involves a large amount of data, deep neural network is used for the detection. Along with the introduction of batch normalization, the deep neural network becomes effective, and also the time taken by the training process is less. Probabilistic neural network (PNN), convolutional neural network (CNN), and recurrent neural network (RNN) are also used for performance analysis and comparison. Deep neural network with batch normalization gives the highest accuracy of 94.35%.

Deep Learning Architectures and Tools

Deep learning is one of the popular machine learning strategies that learns in a supervised or unsupervised manner by forming a cascade of multiple layers of non-linear processing units. It is inspired by the way of information processing and communication pattern of the typical biological nervous system. The deep learning algorithms learn through multiple levels of abstractions and hierarchy of concepts; as a result, it is found to be more efficient than the conventional non-deep machine learning algorithms. This chapter explains the basics of deep learning by highlighting the necessity of deep learning over non-deep learning. It also covers discussion on several recently developed deep learning architectures and popular tools available in market for deep learning, which includes Tensorflow, PyTorch, Keras, Caffe, Deeplearning4j, Pylearn2, Theano, CuDDN, CUDA-Convnet, and Matlab.

Liver Disease Detection Using Grey Wolf Optimization and Random Forest Classification

Utilizing machine learning approaches as non-obtrusive strategies is an elective technique in organizing perpetual liver infections for staying away from the downsides of biopsy. This chapter assesses diverse machine learning methods in expectation of cutting-edge fibrosis by joining the serum bio-markers and clinical data to build up the order models. An imminent accomplice of patients with incessant hepatitis C was separated into two sets—one classified as gentle to direct fibrosis (F0-F2) and the other ordered as cutting-edge fibrosis (F3-F4) as per METAVIR score. Grey wolf optimization, random forest classifier, and decision tree procedure models for cutting-edge fibrosis chance expectation were created. Recipient working trademark bend investigation was performed to assess the execution of the proposed models.

Deep Learning in Engineering Education

The goodness measure of any institute lies in minimising the dropouts and targeting good placements. So, predicting students' performance is very interesting and an important task for educational information systems. Machine learning and deep learning are the emerging areas that truly entice more research practices. This research focuses on applying the deep learning methods to educational data for classification and prediction. The educational data of students from engineering domain with cognitive and non-cognitive parameters is considered. The hybrid model with support vector machine (SVM) and deep belief network (DBN) is devised. The SVM predicts class labels from preprocessed data. These class labels and actual class labels act as input to the DBN to perform final classification. The hybrid model is further optimised using cuckoo search with levy flight. The results clearly show that the proposed model SVM-LCDBN gives better performance as compared to simple hybrid model and hybrid model with traditional cuckoo search.

Analysis of Heart Disorder by Using Machine Learning Methods and Data Mining Techniques

Data mining is the most famous knowledge extraction approach for knowledge discovery from data (KDD). Machine learning is used to enable a program to analyze data, recognize correlations, and make usage on insights to solve issues and/or enrich data and because of prediction. The chapter highlights the need for more research within the usage of robust data mining methods in imitation of help healthcare specialists between the diagnosis regarding heart diseases and other debilitating disease conditions. Heart disease is the primary reason of death of people in the world. Nearly 47% of death is caused by heart disease. The authors use algorithms including random forest, naïve Bayes, support vector machine to analyze heart disease. Accuracy on the prediction stage is high when using a greater number of attributes. The goal is to function predictive evaluation using data mining, using data mining to analyze heart disease, and show which methods are effective and efficient.

A Fully Automated Crop Disease Monitoring and Management System Based on IoT

In recent years, the IoT has evolved and plays a significant role in many fields like smart city, precision farm, traffic signal control system, and so on. In this chapter, an IoT-based crop disease management (CDM) system is proposed that adopts statistical methods for identifying disease, recognizing a right pesticide, and recommending a right pesticide to farmers. The proposed CDM system monitors the agricultural crops with the help of a CCD camera. The camera continuously photographs the crops and sends them to a Raspberry PI processor, which is placed at a workstation and it is connected to the camera with the help of IoT components. The proposed CDM system analyses the crop leaf images, such as removes noise; segments region of interest (RoI), that is, diseased part of the leaf image; extracts features from the RoI; and identifies the disease and takes appropriate measures to control the disease. The proposed IoT-based CDM system was experimented, and the results obtained encourage both the farmers and the researchers in this field.