Fig 3 - uploaded by Mehdi Mohammadi
Content may be subject to copyright.
Google Trend showing more attention toward deep learning in recent years. 

Google Trend showing more attention toward deep learning in recent years. 

Source publication
Article
Full-text available
In the era of the Internet of Things (IoT), an enormous amount of sensing devices collect and/or generate various sensory data over time for a wide range of fields and applications. Based on the nature of the application, these devices will result in big or fast/real-time data streams. Applying analytics over such data streams to discover new infor...

Similar publications

Article
Full-text available
Present digital world with the concept of Smart cities are producing enormous amount of data over the period of time at different sectors of life. It is becoming difficult to track information and produce knowledge out of it without using emerging information management technologies. Big Data Analytics is a must to make use of data generated by IoT...
Preprint
Full-text available
The Internet of Things (IoT) envisions a world-wide, interconnected network of smart physical entities. These physical entities generate a large amount of data in operation and as the IoT gains momentum in terms of deployment, the combined scale of those data seems destined to continue to grow. Increasingly, applications for the IoT involve analyti...

Citations

... On the other hand, the deployment of AI big data analytics is perceived as crucial in understanding the cycle of big data or the procedures that form part of the retrieval, storage, or aggregation of datasets that are diverse and obtained from different sources [168]. Such processes may include the management of risk, goal management, action management, management of decision, management of context, pattern management, or even management of data [169]. Therefore, to a larger extent, these findings are integral in depicting the utility of AI in the assessment of datasets that are connected to simply understand by deploying a language processing approach, which is natural. ...
Chapter
The chapter comprises six sections. Following the introduction is the intelligent or smart sensors and IoT section, which discusses sensing components, processing methods, communication solutions, acting components, nature-inspired approaches, and energy footprint. The third section is the application of smart sensors in robotics, which discusses robotics and smart sensors, improving the perception of the surrounding environment, robotic solutions for better acting, and sensors for human-machine interaction. The fourth section is about the applications of smart systems for wellbeing, which discusses applications of intelligent sensors in agriculture, applications of intelligent sensors in telemedicine, malfunction detection, operation health monitoring, patient assisting mechanisms, and other smart system applications. The fifth section outlines future perspectives and research needs. The final section is dedicated to conclusions.
... For example, AI-driven analytics enable real-time data processing, providing actionable decision-making insights (Yin & Kaynak, 2015). Simultaneously, IoT connects devices within business ecosystems, facilitating seamless workflows and minimizing human error (Mohammadi et al., 2018). Cloud computing further enhances scalability and operational flexibility, enabling businesses to manage resources more effectively. ...
Article
Full-text available
    Purpose: This study examines the challenges and strategies of adopting new technologies to enhance organizational growth and competitiveness in the digital era. It focuses on identifying financial, skill-based, cultural, and external barriers and exploring practical approaches to overcoming them. Research Design and Methodology: A qualitative systematic literature review (SLR) method was employed, synthesizing findings from peer-reviewed articles and academic sources. This approach allowed for a comprehensive analysis of empirical and theoretical perspectives, integrating concepts from innovation diffusion, financial management, and organizational change theories. Findings and Discussion: The study identifies financial constraints, skill gaps, resistance to change, and external factors such as regulatory challenges and inadequate digital infrastructure as critical barriers to technology adoption. It highlights the dynamic interactions among these barriers, revealing how they collectively influence organizational readiness for digital transformation. Strategies such as pilot testing, portfolio diversification, continuous employee training, and regulatory alignment are discussed as essential for overcoming these challenges. The findings also emphasize the role of advanced technologies like Artificial Intelligence (AI), the Internet of Things (IoT), and cloud computing in driving operational efficiency, innovation, and market competitiveness. Implications: The research contributes to academic discourse by offering a holistic framework for understanding technology adoption challenges. Practically, it provides actionable insights for organizations to foster a culture of innovation and align technological initiatives with business goals. Policymakers are urged to invest in digital infrastructure and create supportive regulations to facilitate sustainable digital transformation.
    ... The rapid advancement of technology has led to unprecedented business growth through mergers and acquisitions (Mohammadi et al., 2018) [1]. This growth necessitates the development of effective optimization techniques to ensure success and sustainability [2,3]. ...
    ... The rapid advancement of technology has led to unprecedented business growth through mergers and acquisitions (Mohammadi et al., 2018) [1]. This growth necessitates the development of effective optimization techniques to ensure success and sustainability [2,3]. ...
    ... Data sampling involved determining and selecting only KPIs that were correlated with other KPIs in the whole dataset. This was achieved by calculating correlation coefficients of the KPIs through Equation (1). KPIs that had negative (<0) or positive (>0) correlations with other KPIs were selected, while those KPIs with correlation values of zero (0) were not selected, as they indicated that they were not correlated with other KPIs in the whole dataset. ...
    Article
    Full-text available
    Key performance indicators (KPIs) are crucial for managing business performance and optimization strategies. However, traditional KPIs are inflexible and cannot adapt to changes in staff, business units, functions, and processes. To address this issue, this paper proposes a method that combines statistics, machine learning (ML), and artificial intelligence (AI) to augment traditional KPIs with the flexibility of data-driven automation (DDA) techniques. This study builds a model that takes traditional KPIs generated by an integrated ecosystem as input data and assesses the suitability and correlation of the data using statistical techniques, such as Bartlett’s test of sphericity and the Kaiser–Meyer–Olkin (KMO) test of sampling adequacy. The model then employs exploratory Factor Analysis (FA) techniques to identify correlations and patterns, prioritize KPIs, and automatically generate smart KPIs for business optimization. The model is designed to adapt automatically by creating new KPIs as the business evolves and data change. A case study evaluation validates this approach, showing that DDA techniques can effectively create smart KPIs for business optimization. This approach provides a flexible and adaptable way to manage business performance and optimization strategies, enabling organizations to stay ahead of the competition and achieve their goals.
    ... With the rise of artificial intelligence (AI) and machine learning (ML), these technologies have significantly advanced the VNE space, providing innovative and sophisticated solutions [314]. ML-based solutions are preferred over classical optimization schemes because they can handle highdimensional data and learn complex patterns without explicit programming. ...
    Article
    Network virtualization (NV) allows service providers (SPs) to instantiate logically isolated entities called virtual networks (VNs) on top of a substrate network (SN). Though VNs bring about multiple benefits, particularly in terms of economic costs and elasticity, they also force various technical challenges to be addressed. The primary one is the issue of optimally allocating resources to VNs, also termed virtual network embedding (VNE). This paper presents an exhaustive survey of VNE by extensively covering the state-of-the-art research field in this very active field and focusing on the emerging research trends in industry and academia over the last decade. In addition, this survey originally contributes to the literature by proposing a novel taxonomy for existing VNE solutions and providing a thorough comparative study of their strategies.
    ... Devices at the edge may convert into fog or may send data to the cloud using networking technologies. The sensed data can be stored and analyzed at the edge, fog, or cloud [12]. ...
    Article
    Full-text available
    In the context of elderly living, occurrence of fall can create critical health deterioration and it is a significant health concerns during their stay in home. As a result, having reliable and timely prediction that pays high attention can help them in their daily life. In this research, the information obtained from a home environment with the assistance of Internet of Technology (IoT) is classified by utilising a feed forward neural network (FFNN). Information enters at the input unit will flows via layer across the network until it reaches the output unit. There is no feedback amongst layers during normal operating conditions, which is when it functions as a classifier. This is why they're referred to as FFNN. A significant element in fall detection is detecting human posture, because fall events are similar to falling posture. A fall can be effectively identified by a distinguishing feature. The proposed system can clearly detect the normal lying posture and lying after falling and it can efficiently detect action motion as well as recognize the action posture. The performance of the proposed and exiting is compared where FFNN achieves 98% detection accuracy, which utilises standard benchmark data sets.
    ... Mohammadi et al. [40] presented a comprehensive overview of the various deep learning architectures and algorithms employed in data analytics within the domain of sensing devices. The publication offers a valuable summary of the most commonly used DL models for sensor data processing, a review of practical DL approaches and use cases, and an identification of the challenges and future research directions in the field of DL for sensing systems. ...
    Article
    Full-text available
    The core of this publication is the design of a system for evaluating the condition of production equipment and machines by monitoring selected parameters of the production process with an additional sensor subsystem. The main positive of the design is the processing of data from the sensor layer using artificial intelligence (AI) and expert systems (ESs) with the use of edge computing (EC). Sensor information is processed directly at the sensor level on the monitored equipment, and the results of the individual subsystems are stored in the form of triggers in a database for use in the predictive maintenance process. The whole solution includes the design of suitable sensors and of the implementation of the sensor layer, the description of data processing algorithms, the design on the communication infrastructure for the whole system, and tests in the form of experimental operation of the device in laboratory conditions. The solution includes the visualisation of the production system status for the operator using an interactive online map.
    ... Deep models include two or more hidden layers and use complex training techniques. Recurrent neural networks with one hidden layer are deep because their units contain a cycle that can be unrolled to a deep network (Mohammadi et al. 2018) . ...
    ... The goal of the authors (Hussain and Park 2021), (Hussain et al. 2022) EEG characteristics will be quantified to better understand task-induced neurological deficits caused by stroke and to analyze biomarkers to identify ischemic stroke patients from healthy people. In training and predicting, DL models often outperform ML techniques in two ways (Mohammadi et al. 2018). They first reduce human training and then delete elements that may be unclear to humans (LeCun et al. 2015). ...
    ... However, the discriminative model, given actual data samples and random noise samples, classifies samples from both sources. After measuring their performance, both models are repeatedly modified such that the discriminative model's output helps the generative model enhance the next iteration's samples (Mohammadi et al. 2018). Many benefits of a GAN include: ...
    Article
    Full-text available
    Due to its widespread perception as a crucial element of the Internet of the future, the Internet of Things (IoT) has garnered a lot of attention in recent years. The Internet of Things (IoT) is made up of billions of sentients, communicative "things" that expand the boundaries of the physical and virtual worlds. Every day, such widely used smart gadgets generate enormous amounts of data, creating an urgent need for rapid data analysis across a range of smart mobile devices. Thankfully, current developments in deep learning have made it possible for us to solve the issue tastefully. Deep models may be built to handle large amounts of sensor data and rapidly and effectively learn underlying properties for a variety of Internet of Things applications on smart devices. We review the research on applying deep learning to several Internet of Things applications in this post. Our goal is to provide insights into the many ways in which deep learning techniques may be used to support Internet of Things applications in four typical domains: smart industrial, smart home, smart healthcare, and smart transportation. One of the main goals is to seamlessly integrate deep learning and IoT, leading to a variety of novel ideas in IoT applications, including autonomous driving, manufacture inspection, intelligent control, indoor localization, health monitoring, disease analysis, and home robotics. We also go over a number of problems, difficulties, and potential avenues for future study that make use of deep learning (DL), which is turning out to be one of the most effective and appropriate methods for dealing with various IoT security concerns. The goal of recent research has been to enhance deep learning algorithms for better Internet of Things security. This study examines deep learning-based intrusion detection techniques, evaluates the effectiveness of several deep learning techniques, and determines the most effective approach for deploying intrusion detection in the Internet of Things. This study uses Deep Learning (DL) approaches to better expand intelligence and application skills by using the large quantity of data generated or acquired. The many IoT domains have drawn the attention of several academics, and both DL and IoT approaches have been explored. Because DL was designed to handle a variety of data in huge volumes and required processing in virtually real-time, it was indicated by several studies as a workable method for handling data generated by IoT.
    ... Para [Mohammadi et al. 2018] os dados de IoT são classificados como de big data, mas são diferentes, sendo necessário explorar as propriedades dos dados de IoT e tais diferenças em relação a dados tradicionais. Algumas características importantes sobre os dados de IoT: são dados de streaming em grande escala; possuem heterogeneidade, ou seja, vários dispositivos de aquisição de dados IoT reúnem informações diferentes; apresentam correlação de tempo e espaço, onde, os dispositivos sensores são conectados a um local específico e, portanto, têm um local e um carimbo de data/hora para cada um dos itens de dados; são dados com alto grau de ruído, devido a pequenos pedaços de dados em aplicativos de IoT, estando sujeito a erros e ruídos durante a aquisição e transmissão. ...
    Book
    Full-text available
    O Programa Clube Universidade Hacker tem como objetivo acompanhar a evolução tecnológica e aprofundar conhecimentos em segurança da informação por meio de diversas iniciativas, incluindo momentos culturais, workshops com especialistas da indústria, oficinas práticas, treinamentos online e presenciais, promoção de eventos técnico-científicos, interação com grupos externos de tecnologia e segurança, desenvolvimento de soluções tecnológicas para demandas locais e regionais, campanhas de conscientização sobre tecnologia, segurança e privacidade, além de competições de hack e outras atividades de integração.
    ... The accuracy of Nox emission is not ever reduced anymore (Wang et al., 2023) In the combustion process, the entity technique is very rarely applied. The development of a CNN soft sensor system for a real-time combustion process quality prediction is attempted for the first time in this research (Mohammadi et al., 2018).Using the entity model, a reduction in the losses and increase in the efficiency can be done. The following topics are discussed in the present work: ...
    ... The experiment is simulated in Python 3.6, and PyTorch is the framework used. PyTorch is a machine learning tool kit built on the Torch library (Habib et al., 2011;Ketkar & Santana, 2017;Mannes, 2017) and widely used by Meta AI (Mohammadi, 2018) for applications like artificial intelligence and natural language processing (Patel, 2017). It is open-source software available for free, as per the Modified BSD license. ...
    Article
    Full-text available
    PyTorch, an open-source machine learning framework built on the Torch library, is used in this application to apply deep learning to image classification in a boiler section and to design an entity algorithm for predicting the amount of oxygen available in the furnace section. The physical features of this flame are viewable using pictures obtained from a Charge Coupled Device (CCD). By removing the nonlinear elements, a multilayer CNN forecasts the amount of oxygen in the flue gas from a boiler. From the results of experiments conducted on-site in a real-time combustion system, images of boilers under various settings, including temperatures, air pressures, and gas conditions, have been obtained. Classification models are then applied. The precise quantity of oxygen content is calculated with these photos as input and comparing the outcomes with the test data set. More insightful information about the flame's physical features can be defined using a Convolutional Neural Network (CNN) model and a multilayer representation of the flame images. The flame images captured on-site from an actual combustion system are utilized to illustrate this notion. The oxygen content is predicted using a multilevel-based, unsupervised, and semi-supervised deep entity algorithm by taking 12 classes and training 4,203,592 images each flame image in the tests has a resolution of 24 bits per pixel and a size of 658*492 pixels. After training the model, the loss is as low as 3%, and the attained accuracy is 97%.
    ... W ITH the rapid development of Industrial Internet of Things (IIoT), modern industrial processes are becoming increasingly automated, interconnected. IIoT systems integrate a multitude of contemporary sensors and controllers linked through advanced network [1] infrastructures to optimize operations, enhance efficiency, and improve decision-making. These sensors and controllers are equipped with intelligent capabilities that produce and send out multidimensional and multi-modal data essential for real-time monitoring and management in industrial settings. ...
    ... Subject (c) and (d) have the highest cosine similarity for normal data reconstruction, concentrated in the [0.8,1] interval. (b) is concentrated in [0.6, 0.9] interval and (a) is in [0.5, 0.8] interval. ...
    Preprint
    Full-text available
    Anomaly detection of multivariate time series (MTS) is crucial in industrial intelligent systems. To address the challenges of absence of anomaly labels, fast inference time, multi-source and multi-modality in anomaly detection, researchers have primarily investigated unsupervised reconstruction-driven methods. However, the existing reconstruction-driven methods mainly focus on minimizing reconstruction errors while neglecting the importance of training methods that increase errors between normal and abnormal classes. Furthermore, accurately constructing the feature space of normal and abnormal classes during the reconstruction process remains a challenge. In this paper, we propose an innovative model, namely the confidence adversarial autoencoder (CAAE). The proposed CAAE combines a confidence network, based on window credibility judgment, with an autoencoder to provide credibility support for anomaly detection. We further introduce fake labels to provide the confidence network with a discriminative knowledge for identifying reconstructed data. Additionally, we implement the confidence adversarial training method to generate fake labels to construct an adversarial loss aiming to expand the decision boundary of anomaly scores. Extensive experimental results on publicly available time series datasets are provided to demonstrate the efficiency of our proposed CAAE. It reveals that excellent generalization ability and superior average performance are achieved on different datasets compared with the state-of-the-art methods. INDEX TERMS Multivariate time series (MTS), anomaly detection, adversarial training, unsupervised learning, autoencoder.