Preface

Chapter 9 **Machine Learning in Educational Technology 175**

Chapter 10 **Sentiment-Based Semantic Rule Learning for Improved Product**

Dandibhotla Teja Santosh and Bulusu Vishnu Vardhan

Chapter 11 **A Multilevel Evolutionary Algorithm Applied to the Maximum**

Noureddine Bouhmala, Kjell Ivar Øvergård and Karina Hjelmervik

Ibtehal Talal Nafea

**VI** Contents

**Recommendations 185**

**Satisfiability Problems 203**

The volume of data that is generated, stored, and communicated across different industrial sections, business units, and scientific research communities has been rapidly expanding. The recent developments in cellular telecommunications and distributed/parallel computation technology have enabled real-time collection and processing of the generated data across dif‐ ferent sections. On the one hand, the internet of things (IoT) enabled by cellular telecommu‐ nication industry connects various types of sensors that can collect heterogeneous data and communicate them to the distributed processing units. On the other hand, the recent advan‐ ces in computational capabilities such as parallel processing in graphical processing units (GPUs) and distributed processing over cloud computing clusters have enabled the process‐ ing of a vast amount of data. There has been a vital need to discover important patterns and infer trends from a large volume of data (so-called Big Data) to empower data-driven deci‐ sion-making processes. Tools and techniques have been developed in machine learning to draw insightful conclusions from available data in a structured and automated fashion. Ma‐ chine learning algorithms are based on concepts and tools developed in several fields includ‐ ing statistics, artificial intelligence, information theory, cognitive science, and control theory. The recent advances in the area of machine learning have had a broad range of applications in different scientific disciplines (e.g., bioinformatics, particle physics, and neuroscience), technology (e.g., autonomous cyber physical systems, and computer-assisted diagnosis sys‐ tems), and business (e.g., online financial trading tools and data-driven marketing).

This book is an attempt to bring together the recent developments in machine learning tools and techniques, and the emerging applications of machine learning across different research disciplines, industry sections, and business areas. The book starts with a chapter on hard‐ ware accelerator design for the implementation of machine learning algorithms. The chapter addresses hardware architectures to enable parallel processing and hence accelerating the execution of these algorithms. The book continues with subsequent chapters, each of which addresses a specific area where machine learning can achieve substantial improvements.

Predictive models for air quality prediction are discussed in Chapter 2. The chapter presents several algorithms that are developed to predict the concentration of particles suspended in the air. The predictions are based on heterogeneous datasets collected with various sensors that measure traffic status, weather condition, and trace gases. The presented techniques would be able to advance public health by providing accurate air quality prediction in fu‐ ture smart cities. Automated event detection is another area where machine learning can improve spreading awareness and organizing responses in emerging smart cities. Automat‐ ed event detection using multimedia information in tweets is addressed in Chapter 3. A multiple kernel learning (MKL) algorithm is presented to automatically combine extracted features from texts and images in tweets and detect events. This would assist risk manage‐ ment strategies, improve public health preparedness, and lead to better disaster manage‐ ment. Chapter 4 presents sentiment analysis and machine learning algorithms to determine citizens' perceptions using tweet messages. The presented techniques enable inferring of public opinion using social media.

and consumer online reviews, can be used to propose the best recommendations to the con‐ sumers. Chapter 10 addresses this topic and presents product review opinion ontology to annotate product features and online product reviews. This can be used to provide product recommendation and to help the consumers to make correct purchase decisions. Chapter 11 presents a combinatorial optimization algorithm to find a feasible solution for a set of con‐

The book provides an overview of machine learning techniques and their potential applica‐ tions in different areas. The book can be used by researchers, business owners, entrepre‐ neurs, and engineers to have an insight into the potential opportunities in developing machine learning solutions in different areas. The book has been the result of the effort of several authors with very diverse technical backgrounds who are experts in different fields. The editor would like to use this opportunity to thank all the authors who have contributed to this book with their valuable inputs to the technical chapters. In addition, the editor would like to thank the publication team at IntechOpen publisher who facilitated the publi‐ cation process. The editor thanks the publication manager Mr. Teo Kos for all his help.

> **Dr. Hamed Farhadi** Experienced Researcher Ericsson Research Stockholm, Sweden

Preface IX

straints in a timely manner.

The developments in scientific disciplines are experiencing a transition from computational‐ ly-intensive to data-intensive problems to be addressed because of the extensive data that are available from scientific experiments/simulations. The investigation of the interactions of laser pulses with plasmas is a rapidly developing area of physics where the need for the development of predictive models based on extensive data is inevitable. Chapter 5 presents a predictive system for laser-plasma interactions that incorporates big data and advanced machine learning algorithms for fast and reliable prediction of the outcome of the interac‐ tions. The predictive system enables the discovery and understanding of various physical phenomena occurring during the interactions, hence, allowing researchers to set up control‐ led experiments at optimal parameters.

Radio communication networks are one of the key enablers of smart cities, and it is expected that a tremendous number of sensors and devices are to be connected through these net‐ works. Network traffic is rapidly increasing and, thus, efficient utilization of the spectrum is of vital importance for the sustainable development of future radio networks. Cognitive ra‐ dios would possibly enhance spectrum utilization by introducing an opportunistic usage of the frequency bands that are not occupied by licensed spectrum users. The detection of active licensed users is a key technical requirement to prevent making interference in the licensed spectrum. Chapter 6 presents a new approach for the detection of licensed users' availability based on machine learning techniques. Indoor positioning systems based on already existing hardware such as Wi-Fi transceivers are gaining momentum. These systems extract features from the received signals and try to estimate the object's location using a codebook based on the received signals at reference points. Thus, the problem can be formulated as a classifica‐ tion problem, which has been discussed in machine learning discipline. In Chapter 7, a locali‐ zation technique inspired by machine learning techniques is proposed. The proposed algorithms can achieve a sufficiently accurate localization precision for indoor navigations.

Automation of the diagnosis of life-threatening diseases such as Malaria would substantially improve public health in resource-scarce areas. Machine learning technologies can be used for automated diagnosis of Malaria. Chapter 8 presents an accurate method for the clas‐ sification of malaria-infected cells using deep convolutional neural networks. The chapter first describes preprocessing methods to be applied on blood cell images. Next, a procedure for compiling a pathologist-curated image dataset for training the deep neural network is presented. The classification accuracies obtained by deep convolutional neural networks through training, validating, and testing with various combinations of datasets including the original dataset and the augmented datasets are discussed in the chapter.

Machine learning can have a profound impact on future education systems by providing a customized learning experience in remote education systems. These systems can be virtual assistants of teachers enabling teachers to have a better understanding concerning how their students are progressing with learning. Chapter 9 addresses possible developments that ma‐ chine learning algorithms can provide in future education systems.

Advanced learning algorithms would facilitate businesses through the development of dataoriented product recommendations. Heterogeneous datasets, including product features and consumer online reviews, can be used to propose the best recommendations to the con‐ sumers. Chapter 10 addresses this topic and presents product review opinion ontology to annotate product features and online product reviews. This can be used to provide product recommendation and to help the consumers to make correct purchase decisions. Chapter 11 presents a combinatorial optimization algorithm to find a feasible solution for a set of con‐ straints in a timely manner.

ment strategies, improve public health preparedness, and lead to better disaster manage‐ ment. Chapter 4 presents sentiment analysis and machine learning algorithms to determine citizens' perceptions using tweet messages. The presented techniques enable inferring of

The developments in scientific disciplines are experiencing a transition from computational‐ ly-intensive to data-intensive problems to be addressed because of the extensive data that are available from scientific experiments/simulations. The investigation of the interactions of laser pulses with plasmas is a rapidly developing area of physics where the need for the development of predictive models based on extensive data is inevitable. Chapter 5 presents a predictive system for laser-plasma interactions that incorporates big data and advanced machine learning algorithms for fast and reliable prediction of the outcome of the interac‐ tions. The predictive system enables the discovery and understanding of various physical phenomena occurring during the interactions, hence, allowing researchers to set up control‐

Radio communication networks are one of the key enablers of smart cities, and it is expected that a tremendous number of sensors and devices are to be connected through these net‐ works. Network traffic is rapidly increasing and, thus, efficient utilization of the spectrum is of vital importance for the sustainable development of future radio networks. Cognitive ra‐ dios would possibly enhance spectrum utilization by introducing an opportunistic usage of the frequency bands that are not occupied by licensed spectrum users. The detection of active licensed users is a key technical requirement to prevent making interference in the licensed spectrum. Chapter 6 presents a new approach for the detection of licensed users' availability based on machine learning techniques. Indoor positioning systems based on already existing hardware such as Wi-Fi transceivers are gaining momentum. These systems extract features from the received signals and try to estimate the object's location using a codebook based on the received signals at reference points. Thus, the problem can be formulated as a classifica‐ tion problem, which has been discussed in machine learning discipline. In Chapter 7, a locali‐ zation technique inspired by machine learning techniques is proposed. The proposed algorithms can achieve a sufficiently accurate localization precision for indoor navigations. Automation of the diagnosis of life-threatening diseases such as Malaria would substantially improve public health in resource-scarce areas. Machine learning technologies can be used for automated diagnosis of Malaria. Chapter 8 presents an accurate method for the clas‐ sification of malaria-infected cells using deep convolutional neural networks. The chapter first describes preprocessing methods to be applied on blood cell images. Next, a procedure for compiling a pathologist-curated image dataset for training the deep neural network is presented. The classification accuracies obtained by deep convolutional neural networks through training, validating, and testing with various combinations of datasets including

the original dataset and the augmented datasets are discussed in the chapter.

chine learning algorithms can provide in future education systems.

Machine learning can have a profound impact on future education systems by providing a customized learning experience in remote education systems. These systems can be virtual assistants of teachers enabling teachers to have a better understanding concerning how their students are progressing with learning. Chapter 9 addresses possible developments that ma‐

Advanced learning algorithms would facilitate businesses through the development of dataoriented product recommendations. Heterogeneous datasets, including product features

public opinion using social media.

VIII Preface

led experiments at optimal parameters.

The book provides an overview of machine learning techniques and their potential applica‐ tions in different areas. The book can be used by researchers, business owners, entrepre‐ neurs, and engineers to have an insight into the potential opportunities in developing machine learning solutions in different areas. The book has been the result of the effort of several authors with very diverse technical backgrounds who are experts in different fields. The editor would like to use this opportunity to thank all the authors who have contributed to this book with their valuable inputs to the technical chapters. In addition, the editor would like to thank the publication team at IntechOpen publisher who facilitated the publi‐ cation process. The editor thanks the publication manager Mr. Teo Kos for all his help.

> **Dr. Hamed Farhadi** Experienced Researcher Ericsson Research Stockholm, Sweden

**Chapter 1**

Provisional chapter

**Hardware Accelerator Design for Machine Learning**

DOI: 10.5772/intechopen.72845

Hardware Accelerator Design for Machine Learning

Machine learning is widely used in many modern artificial intelligence applications. Various hardware platforms are implemented to support such applications. Among them, graphics processing unit (GPU) is the most widely used one due to its fast computation speed and compatibility with various algorithms. Field programmable gate arrays (FPGA) show better energy efficiency compared with GPU when computing machine learning algorithm at the cost of low speed. Finally, various application specific integrated circuit (ASIC) architecture is proposed to achieve the best energy efficiency at the cost of less reconfigurability which makes it suitable for special kinds of machine learning algorithms such as a deep convolutional neural network. Finally, analog computing shows a promising methodology to compute large-sized machine learning algorithm due to its low design cost and fast computing speed; however, due to the requirement of the analog-to-digital converter (ADC) in the analog computing, this kind of technique is only applicable to low computation resolu-

tion, making it unsuitable for most artificial intelligence (AI) applications.

Keywords: machine learning, hardware accelerator, model compression, analog

Machine learning (ML) is currently widely used in many modern artificial intelligence (AI) applications [1]. The breakthrough of the computation ability has enabled the system to compute complicated different ML algorithm in a relatively short time, providing real-time human-machine interaction such as face detection for video surveillance, advanced driverassistance systems (ADAS), and image recognition early cancer detection [2, 3]. Among all those applications, a high detection accuracy requires complicated ML computation, which comes at the cost of high computational complexity. This results in a high requirement on the hardware platform. Currently, most applications are implemented on general-purpose compute engines, especially graphics processing units (GPUs). However, work recently reported

> © The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and eproduction in any medium, provided the original work is properly cited.

distribution, and reproduction in any medium, provided the original work is properly cited.

© 2018 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,

Li Du and Yuan Du

Li Du and Yuan Du

Abstract

1. Introduction

Additional information is available at the end of the chapter

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/intechopen.72845

computing, GPU, FPGA, ASIC

Provisional chapter
