More Web Proxy on the site http://driver.im/

You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.

Search for Articles:

Title / Keyword

Author / Affiliation / Email

Journal

Article Type

Advanced Search

Section

Special Issue

Volume

Issue

Number

Page

Logical OperatorOperator

Search Text

Search Type

Rain-Free Vision: Robust Image Deraining for Enhanced Autonomous Driving Safety
Area–Time-Efficient High-Radix Modular Inversion Algorithm and Hardware Implementation for ECC over Prime Fields
The Use of eXplainable Artificial Intelligence and Machine Learning Operation Principles to Support the Continuous Development of Machine Learning-Based Solutions in Fault Detection and Identification
Creating a Newer and Improved Procedural Content Generation (PCG) Algorithm with Minimal Human Intervention for Computer Gaming Development

Journal Description

Computers

Computers is an international, scientific, peer-reviewed, open access journal of computer science, including computer and network architecture and computer–human interaction as its main foci, published monthly online by MDPI.

Open Access— free for readers, with article processing charges (APC) paid by authors or their institutions.
High Visibility: indexed within Scopus, ESCI (Web of Science), dblp, Inspec, and other databases.
Journal Rank: JCR - Q2 (Computer Science, Interdisciplinary Applications) / CiteScore - Q2 (Computer Networks and Communications)
Rapid Publication: manuscripts are peer-reviewed and a first decision is provided to authors approximately 15.5 days after submission; acceptance to publication is undertaken in 3.8 days (median values for papers published in this journal in the second half of 2024).
Recognition of Reviewers: reviewers who provide timely, thorough peer-review reports receive vouchers entitling them to a discount on the APC of their next publication in any MDPI journal, in appreciation of the work done.

Impact Factor: 2.6 (2023); 5-Year Impact Factor: 2.4 (2023)

Imprint Information Journal Flyer Open Access ISSN: 2073-431X

Latest Articles

19 pages, 529 KiB

Open AccessReview

Redefining Event Detection and Information Dissemination: Lessons from X (Twitter) Data Streams and Beyond

by Harshit Srivastava and Ravi Sankar

Computers 2025, 14(2), 42; https://doi.org/10.3390/computers14020042 - 28 Jan 2025

X (formerly known as Twitter), Reddit, and other social media forums have dramatically changed the way society interacts with live events in this day and age. The huge amount of data generated by these platforms presents challenges, especially in terms of processing speed [...] Read more.

X (formerly known as Twitter), Reddit, and other social media forums have dramatically changed the way society interacts with live events in this day and age. The huge amount of data generated by these platforms presents challenges, especially in terms of processing speed and the complexity of finding meaningful patterns and events. These data streams are generated in multiple formats, with constant updating, and are real-time in nature; thus, they require sophisticated algorithms capable of dynamic event detection in this dynamic environment. Event detection techniques have recently achieved substantial development, but most research carried out so far evaluates only single methods, not comparing the overall performance of these methods across multiple platforms and types of data. With that view, this paper represents a deep investigation of complex state-of-the-art event detection algorithms specifically customized for streams of data from X. We review various current techniques based on a thorough comparative performance test and point to problems inherently related to the detection of patterns in high-velocity streams with noise. We introduce some novelty to this research area, supported by appropriate robust experimental frameworks, to performed comparisons quantitatively and qualitatively. We provide insight into how those algorithms perform under varying conditions by defining a set of clear, measurable metrics. Our findings contribute new knowledge that will help inform future research into the improvement of event detection systems for dynamic data streams and enhance their capabilities for real-time and actionable insights. This paper will go a step further than the present knowledge of event detection and discuss how algorithms can be adapted and refined in view of the emerging demands imposed by data streams. Full article

(This article belongs to the Special Issue Recent Advances in Social Networks and Social Media)

► Show Figures

Figure 1

28 pages, 991 KiB

Open AccessReview

A Review of Vessel Time of Arrival Prediction on Waterway Networks: Current Trends, Open Issues, and Future Directions

by Abdullah Al Noman, Aaron Heuermann, Stefan Wiesner and Klaus-Dieter Thoben

Computers 2025, 14(2), 41; https://doi.org/10.3390/computers14020041 - 28 Jan 2025

With the vast majority of global trade volume and value reliant on maritime transport, accurate prediction of vessel estimated time of arrival (ETA) is crucial for optimizing supply chain efficiency and managing logistical complexities in port operations. This review paper systematically examines the [...] Read more.

With the vast majority of global trade volume and value reliant on maritime transport, accurate prediction of vessel estimated time of arrival (ETA) is crucial for optimizing supply chain efficiency and managing logistical complexities in port operations. This review paper systematically examines the current state of research and practices in the field of vessel ETA prediction, highlighting significant trends, methodologies, and technologies. It explores various approaches, including classical methods, machine learning and deep learning algorithms, and hybrid methods, developed to enhance the accuracy and reliability of vessel travel time and arrival time predictions. Additionally, this paper categorizes key influencing factors and metrics, and identifies open issues and challenges within current prediction models. Concluding with proposed future research directions aimed at addressing the identified gaps and leveraging technological advancements, this review emphasizes the importance of fostering innovation in maritime ETA prediction systems, particularly within the framework of Intelligent Transportation Systems (ITSs) and maritime logistics. By applying a systematic literature review (SLR) methodology and conducting an in-depth evaluation, the results provide a comprehensive overview of vessel ETA prediction for researchers, practitioners, and policy makers involved in maritime transport and logistics, and offer insights into the potential for improved efficiency, safety, and environmental sustainability in waterway networks. Full article

(This article belongs to the Special Issue IT in Production and Logistics)

20 pages, 3245 KiB

Open AccessReview

Model Transformations Used in IT Project Initial Phases: Systematic Literature Review

by Oksana Nikiforova, Kristaps Babris, Uldis Karlovs-Karlovskis, Marta Narigina, Andrejs Romanovs, Anita Jansone, Janis Grabis and Oscar Pastor

Computers 2025, 14(2), 40; https://doi.org/10.3390/computers14020040 - 27 Jan 2025

The paper emphasizes the critical importance of the initial phase in IT project development to avoid implementation errors. It argues that minimizing these errors can be achieved by developing project artifacts at the early stage using a model-driven engineering-based approach. Model transformation plays [...] Read more.

The paper emphasizes the critical importance of the initial phase in IT project development to avoid implementation errors. It argues that minimizing these errors can be achieved by developing project artifacts at the early stage using a model-driven engineering-based approach. Model transformation plays a basic role in that context. The goal of this paper is to survey publications in which the authors propose generating initial project elements through model-driven engineering and to analyze the level of model transformations offered in their solutions. As a result, the authors would highlight the necessity of understanding which elements of a project can be obtained through automatic transformations and which still require manual manipulation. This distinction is crucial, as it can significantly influence the efficiency and accuracy of the project’s early phases. In general, identifying the project components that can be reliably generated through model transformations helps streamline the project inception and elaboration process performed before IT product implementation. Full article

► Show Figures

Figure 1

Figure 1
<p>Conceptual model for model transformation within model-driven engineering.</p> Full article ">Figure 2
<p>IT project initial stage specification as a scope of the research.</p> Full article ">Figure 3
<p>PRISMA 2020 flow diagram for the literature search process.</p> Full article ">Figure 4
<p>Dispersion of papers published per year.</p> Full article ">Figure 5
<p>Notations used for the source model in the scope of the research.</p> Full article ">Figure 6
<p>Notations used for the target model in the scope of the research.</p> Full article ">Figure 7
<p>Numbers of papers on the transformations between artifacts from different project stages.</p> Full article ">Figure 8
<p>Detailed depiction of all the smaller artifacts (A1–A4) used in the initial stages of the IT project.</p> Full article ">

16 pages, 2717 KiB

Open AccessArticle

Identification of Greek Orthodox Church Chants Using Fuzzy Entropy

by Lazaros Moysis, Konstantinos Karasavvidis, Dimitris Kampelopoulos, Achilles D. Boursianis, Sotirios Sotiroudis, Spiridon Nikolaidis, Christos Volos, Panagiotis Sarigiannidis, Mohammad Abdul Matin and Sotirios K. Goudos

Computers 2025, 14(2), 39; https://doi.org/10.3390/computers14020039 - 27 Jan 2025

In this work, a comparison of Greek Orthodox religious chants is performed using fuzzy entropy. Using a dataset of chant performances, each recitation is segmented into overlapping time windows, and the fuzzy entropy of each window in the frequency domain is computed. We [...] Read more.

In this work, a comparison of Greek Orthodox religious chants is performed using fuzzy entropy. Using a dataset of chant performances, each recitation is segmented into overlapping time windows, and the fuzzy entropy of each window in the frequency domain is computed. We introduce a novel audio fingerprinting framework by comparing the variations in the resulting fuzzy entropy vector for the dataset. For this purpose, we use the correlation coefficient as a measure and dynamic time warping. Thus, it is possible to match the performances of the same chant with high probability. The proposed methodology provides a foundation for building an audio fingerprinting method based on fuzzy entropy. Full article

15 pages, 1916 KiB

Open AccessArticle

Cybercrime Resilience in the Era of Advanced Technologies: Evidence from the Financial Sector of a Developing Country

by Adeel Ali, Mahmood Shah, Monika Foster and Mansour Naser Alraja

Computers 2025, 14(2), 38; https://doi.org/10.3390/computers14020038 - 27 Jan 2025

Technological advancements have helped all sectors to evolve. This advancement has widened the cyberspace and attack surface, which has led to a drastic increase in cyberattacks. Cybersecurity solutions have also evolved. The advancement is relatively slower in developing countries. However, the financial sector [...] Read more.

Technological advancements have helped all sectors to evolve. This advancement has widened the cyberspace and attack surface, which has led to a drastic increase in cyberattacks. Cybersecurity solutions have also evolved. The advancement is relatively slower in developing countries. However, the financial sector in developing countries has shown resistance to cyberattacks. This paper investigates the reasons for this resistance. Despite using legacy systems, the banking sector in Pakistan has demonstrated resistance to cyberattacks. The research used a qualitative approach. Semi-structured interviews were conducted with nine cybersecurity experts in the banking sector to illustrate the reasons for this cybersecurity resistance. The research focused on cybersecurity experts in the banking sector, recognizing that this industry is particularly prone to cyberattacks on a global scale. The study utilised a thematic analysis technique to find resistance factors. The analysis suggests that the opportunity cost of cyberattacks and lower attack surface in developing countries like Pakistan are the main reasons for the lower financial losses. The findings of this research will encourage the adoption of advanced technologies such as artificial intelligence (AI) and machine learning (ML) for cybersecurity in developing countries’ banking and financial sectors. Full article

(This article belongs to the Special Issue Cyber Security and Privacy in IoT Era)

► Show Figures

Figure 1

26 pages, 11614 KiB

Open AccessArticle

Commutative Quaternion Algebra with Quaternion Fourier Transform-Based Alpha-Rooting Color Image Enhancement

by Artyom M. Grigoryan and Alexis A. Gomez

Computers 2025, 14(2), 37; https://doi.org/10.3390/computers14020037 - 26 Jan 2025

In this paper, we describe the associative and commutative algebra or the (2,2)-model of quaternions with application in color image enhancement. The method of alpha-rooting, which is based on the 2D quaternion discrete Fourier transform (QDFT) is considered. In the (2,2)-model, the aperiodic [...] Read more.

In this paper, we describe the associative and commutative algebra or the (2,2)-model of quaternions with application in color image enhancement. The method of alpha-rooting, which is based on the 2D quaternion discrete Fourier transform (QDFT) is considered. In the (2,2)-model, the aperiodic convolution of quaternion signals can be calculated by the product of their QDFTs. The concept of linear convolution is simple, that is, it is unique, and the reduction of this operation to the multiplication in the frequency domain makes this model very attractive for processing color images. Note that in the traditional quaternion algebra, which is not commutative, the convolution can be chosen in many different ways, and the number of possible QDFTs is infinite. And most importantly, the main property of the traditional Fourier transform that states that the aperiodic convolution is the product of the transform in the frequency domain is not valid. We describe the main property of the (2,2)-model of quaternions, the quaternion exponential functions and convolution. Three methods of alpha-rooting based on the 2D QDFT are presented, and illustrative examples on color image enhancement are given. The image enhancement measures to estimate the quality of the color images are described. Examples of the alpha-rooting enhancement on different color images are given and analyzed with the known histogram equalization and Retinex algorithms. Our experimental results show that the alpha-rooting method in the quaternion space is one of the most effective methods of color image enhancement. Quaternions allow all colors in each pixel to be processed as a whole, rather than individually as is done in traditional processing methods. Full article

(This article belongs to the Special Issue Advanced Image Processing and Computer Vision)

34 pages, 1690 KiB

Open AccessReview

Machine Learning Approaches for Speech-Based Alzheimer’s Detection: A Comprehensive Survey

by Ahmed Sharafeldeen, Justin Keowen and Ahmed Shaffie

Computers 2025, 14(2), 36; https://doi.org/10.3390/computers14020036 - 24 Jan 2025

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder that significantly impairs cognitive functions, leading to memory loss and other behavioral changes. It is the seventh leading cause of death worldwide, with millions of people affected. Early and accurate detection of AD is critical [...] Read more.

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder that significantly impairs cognitive functions, leading to memory loss and other behavioral changes. It is the seventh leading cause of death worldwide, with millions of people affected. Early and accurate detection of AD is critical for improving patient outcomes and slowing disease progression. Recent advancements in machine learning (ML) and deep learning (DL) models have demonstrated significant potential for detecting AD using patient’s speech signals, as subtle changes in speech patterns, such as reduced fluency, pronunciation difficulties, and cognitive decline, can serve as early indicators of the disease, offering a non-invasive and cost-effective method for early diagnosis. This survey paper provides a comprehensive review of the current literature on the application of ML and DL techniques for AD detection through the analysis of a patient’s speech signal, utilizing various acoustic and textual features. Moreover, it offers an overview of the changes in the brain caused by the disease, associated risk factors, publicly available datasets, and future directions for leveraging ML and DL in the detection of AD. Full article

► Show Figures

Figure 1

Figure 1
<p>Healthy cadaver brain compared to a cadaver brain with Alzheimer’s disease pathology [<a href="#B10-computers-14-00036" class="html-bibr">10</a>]. Arrowheads indicate sulcal widening and gyrus narrowing in the brain of an Alzheimer’s patient compared to a normal brain, with these features being more pronounced in coronal sections. Arrows highlight the enlargement of the frontal and temporal horns of the lateral ventricles. The circle marks the loss of pigmented neurons in the locus coeruleus within the pontine tegmentum. These changes are not exclusive to AD.</p> Full article ">Figure 2
<p>Progression of Alzheimer’s disease pathology through the brain, illustrating the spread of amyloid plaques and neurofibrillary tangles.</p> Full article ">Figure 3
<p>A visual overview of the CAD system used to differentiate healthy individuals from those with AD or MCI through the analysis of speech signals.</p> Full article ">Figure 4
<p>A bar chart visualizing the performance of literature reviews that employed traditional machine learning models to analyze speech signal to distinguish between healthy individuals and those with AD [<a href="#B27-computers-14-00036" class="html-bibr">27</a>,<a href="#B30-computers-14-00036" class="html-bibr">30</a>,<a href="#B31-computers-14-00036" class="html-bibr">31</a>,<a href="#B33-computers-14-00036" class="html-bibr">33</a>,<a href="#B39-computers-14-00036" class="html-bibr">39</a>,<a href="#B40-computers-14-00036" class="html-bibr">40</a>,<a href="#B41-computers-14-00036" class="html-bibr">41</a>,<a href="#B42-computers-14-00036" class="html-bibr">42</a>,<a href="#B43-computers-14-00036" class="html-bibr">43</a>,<a href="#B47-computers-14-00036" class="html-bibr">47</a>,<a href="#B54-computers-14-00036" class="html-bibr">54</a>,<a href="#B55-computers-14-00036" class="html-bibr">55</a>,<a href="#B56-computers-14-00036" class="html-bibr">56</a>,<a href="#B57-computers-14-00036" class="html-bibr">57</a>,<a href="#B58-computers-14-00036" class="html-bibr">58</a>].</p> Full article ">Figure 5
<p>A bar chart illustrating the effectiveness of literature reviews that utilized deep learning methods to analyze speech signal to differentiate between healthy individuals and those with AD [<a href="#B64-computers-14-00036" class="html-bibr">64</a>,<a href="#B66-computers-14-00036" class="html-bibr">66</a>,<a href="#B74-computers-14-00036" class="html-bibr">74</a>,<a href="#B78-computers-14-00036" class="html-bibr">78</a>,<a href="#B82-computers-14-00036" class="html-bibr">82</a>,<a href="#B83-computers-14-00036" class="html-bibr">83</a>,<a href="#B89-computers-14-00036" class="html-bibr">89</a>,<a href="#B90-computers-14-00036" class="html-bibr">90</a>,<a href="#B91-computers-14-00036" class="html-bibr">91</a>,<a href="#B93-computers-14-00036" class="html-bibr">93</a>,<a href="#B95-computers-14-00036" class="html-bibr">95</a>,<a href="#B98-computers-14-00036" class="html-bibr">98</a>,<a href="#B99-computers-14-00036" class="html-bibr">99</a>,<a href="#B104-computers-14-00036" class="html-bibr">104</a>,<a href="#B106-computers-14-00036" class="html-bibr">106</a>,<a href="#B108-computers-14-00036" class="html-bibr">108</a>].</p> Full article ">

29 pages, 8212 KiB

Open AccessArticle

ApproxGeoMap: An Efficient System for Generating Approximate Geo-Maps from Big Geospatial Data with Quality of Service Guarantees

by Reem Abdelaziz Alshamsi, Isam Mashhour Al Jawarneh, Luca Foschini and Antonio Corradi

Computers 2025, 14(2), 35; https://doi.org/10.3390/computers14020035 - 23 Jan 2025

Timely, region-based geo-maps like choropleths are essential for smart city applications like traffic monitoring and urban planning because they can reveal statistical patterns in geotagged data. However, because data overloading is brought on by the quick inflow of massive geospatial data, creating these [...] Read more.

Timely, region-based geo-maps like choropleths are essential for smart city applications like traffic monitoring and urban planning because they can reveal statistical patterns in geotagged data. However, because data overloading is brought on by the quick inflow of massive geospatial data, creating these visualizations in real time presents serious difficulties. This paper introduces ApproxGeoMap, a novel system designed to efficiently generate approximate geo-maps from fast-arriving georeferenced data streams. ApproxGeoMap employs a stratified spatial sampling method, leveraging geohash tessellation and Earth Mover’s Distance (EMD) to maintain both accuracy and processing speed. We developed a prototype system and tested it on real-world smart city datasets, demonstrating that ApproxGeoMap meets time-based and accuracy-based quality of service (QoS) constraints. Results indicate that ApproxGeoMap significantly enhances efficiency in both running time and map accuracy, offering a reliable solution for high-speed data environments where traditional methods fall short. Full article

(This article belongs to the Special Issue Feature Papers in Computers 2024)

► Show Figures

Figure 1

Figure 1
<p>Region-based geo-maps: Boston, USA. (<b>A</b>) Heatmap of air quality data in Boston, USA. The intensity of the colors (from yellow to red) represents the density of records collected, with red areas indicating the highest density. (<b>B</b>) Choropleth map of air quality data in Boston, USA. Neighborhoods or polygons shaded in darker green represent areas with the highest density of records collected, while lighter colors indicate lower densities.</p> Full article ">Figure 2
<p>Region-based geo-maps: Rome, Italy. (<b>A</b>) Heatmap depicting taxi pickup patterns over one day in Rome, Italy. The color intensity (from yellow to red) indicates the density of taxi pickups, with red areas representing the highest concentration. (<b>B</b>) Choropleth map illustrating the same taxi pickup data in Rome, Italy. The polygons are shaded to reflect the density of taxi pickups, with darker colors representing higher densities and lighter colors indicating lower densities.</p> Full article ">Figure 3
<p>(<b>A</b>,<b>B</b>) Geohash tessellation for Boston, USA. (<b>C</b>,<b>D</b>) Geohash tessellation for Rome, Italy. Geohash tessellation for Boston, USA (<b>A</b>,<b>B</b>), and Rome, Italy (<b>C</b>,<b>D</b>). (<b>A</b>,<b>C</b>) show geohash coverings at precision level 6, characterized by smaller, more granular grid cells, while (<b>B</b>,<b>D</b>) illustrate coverings at precision level 5, with larger, coarser grid cells. Colors represent distinct geohash values within each precision level, highlighting the spatial distribution of data points. Precision corresponds to the number of characters in the geohash value (e.g., sr2yk0 at precision 6 in (<b>C</b>), and sr2yk at precision 5 in (<b>D</b>)), where higher precision offers finer spatial detail, and lower precision aggregates larger spatial areas.</p> Full article ">Figure 4
<p>Earth Movers’ Distance explanation. Visualization of Earth Mover’s Distance (EMD), illustrating the “work” required to transform one distribution into another by moving mass (dust) from piles to fill holes, with cost proportional to the amount moved and the distance covered.</p> Full article ">Figure 5
<p>ApproxGeoMap architecture. This figure illustrates the main components and workflow of the ApproxGeoMap system. The process begins with geohash-based tessellation of input data, followed by stratified-like sampling using geohash regions. The system then aggregates geospatial statistics and visualizes the data through choropleth or heatmaps. The colors in the sample-based choropleth map represent different data densities or attribute values, with darker colors indicating higher values and lighter colors representing lower values. A feedback loop evaluates the error using Earth Mover’s Distance (EMD) to adjust sampling rates and ensure accuracy.</p> Full article ">Figure 6
<p>AGMSampler architecture. The AGMSampler implements stratified-like sampling by iterating through geohash regions within distinct zones (Zone A and Zone B) and selecting data points based on a predefined sampling fraction. Discretization divides these zones into grid cells, where G1 and G2 belong to Zone A but have different geohash values, and G3 and G4 belong to Zone B with distinct geohash values. This process ensures proportional sampling across regions, facilitating accurate and efficient proxy-based aggregation for geospatial data.</p> Full article ">Figure 7
<p>Exploratory Data Analysis: Boston, USA, air quality dataset. The histograms depict temperature, humidity, and pollutant concentrations (NO<sub>2</sub> and PM2.5), highlighting their distributions and frequency across the dataset. The pie chart illustrates the record counts by month, showing the temporal distribution of data collection.</p> Full article ">Figure 8
<p>Exploratory Data Analysis: Rome, Italy, mobility dataset. The bar charts represent driver activity (top 10 most active drivers), data distribution by the month, day of the week, and hour of the day, providing insights into urban mobility trends and temporal activity patterns.</p> Full article ">Figure 9
<p>Comparison of EMD for stratified and random sampling across different sample sizes in Boston, USA. Comparison of Earth Mover’s Distance (lower better) across varying sample sizes for Random Sampling vs. AGM (Stratified Sampling). (<b>A</b>) Results with Geohash Precision Level 5, (<b>B</b>) Results with Geohash Precision Level 6.</p> Full article ">Figure 10
<p>Comparison of RMSE for stratified and random sampling across different sample sizes in Boston, USA. Comparison of RMSE across varying sample sizes for Random Sampling vs. AGM (Stratified Sampling). (<b>A</b>) Results with Geohash Precision Level 5, (<b>B</b>) Results with Geohash Precision Level 6.</p> Full article ">Figure 11
<p>Comparison of EMD for stratified and random sampling across different sample sizes in Rome, Italy. Rome Data: Comparison of Earth Mover’s Distance (lower better) across sample sizes for Random Sampling vs. AGM (Stratified Sampling). (<b>A</b>) Results with Geohash Precision Level 5, (<b>B</b>) Results with Geohash Precision Level 6.</p> Full article ">Figure 12
<p>Comparison of RMSE for stratified and random sampling across different sample sizes in Rome, Italy. Rome Data: Comparison of RMSE across varying sample sizes for Random Sampling vs. AGM (Stratified Sampling). (<b>A</b>) Results with Geohash Precision Level 5, (<b>B</b>) Results with Geohash Precision Level 6.</p> Full article ">Figure 13
<p>Comparison of original, random sampling, and AGM sampling maps for Boston using geohash precision level 6. The figure showcases density variations in Boston using geohash precision level 6. Black regions indicate the highest density of data points, followed by dark blue (moderately high), light blue (moderate), green (low), and white (lowest or no data). The top left map represents the original Boston dataset, duplicated in the bottom left for consistency. The top right map applies random sampling, which introduces disruptions in spatial patterns, while the bottom right map employs ApproxGeoMap (AGM) sampling, demonstrating superior preservation of density patterns and spatial coherence.</p> Full article ">

31 pages, 8973 KiB

Open AccessArticle

Number Recognition Through Color Distortion Using Convolutional Neural Networks

by Christopher Henshaw, Jacob Dennis, Jonathan Nadzam and Alan J. Michaels

Computers 2025, 14(2), 34; https://doi.org/10.3390/computers14020034 - 22 Jan 2025

Machine learning applied to image-based number recognition has made significant strides in recent years. Recent use of Large Language Models (LLMs) in natural language search and generation of text have improved performance for general images, yet performance limitations still exist for data subsets [...] Read more.

Machine learning applied to image-based number recognition has made significant strides in recent years. Recent use of Large Language Models (LLMs) in natural language search and generation of text have improved performance for general images, yet performance limitations still exist for data subsets related to color blindness. In this paper, we replicated the training of six distinct neural networks (MNIST, LeNet5, VGG16, AlexNet, and two AlexNet modifications) using deep learning techniques with the MNIST dataset and the Ishihara-Like MNIST dataset. While many prior works have dealt with MNIST, the Ishihara adaption addresses red-green combinations of color blindness, allowing for further research in color distortion. Through this research, we applied pre-processing to accentuate the effects of red-green and monochrome colorblindness and hyper-parameterized the existing architectures, ultimately achieving better overall performance than currently published in known works. Full article

(This article belongs to the Special Issue Advanced Image Processing and Computer Vision)

► Show Figures

Figure 1

Figure 1
<p>Standard Ishihara Circle [<a href="#B10-computers-14-00034" class="html-bibr">10</a>].</p> Full article ">Figure 2
<p>MNIST Ishihara Circle [<a href="#B8-computers-14-00034" class="html-bibr">8</a>].</p> Full article ">Figure 3
<p>Color Blindness Spectrum [<a href="#B11-computers-14-00034" class="html-bibr">11</a>,<a href="#B15-computers-14-00034" class="html-bibr">15</a>,<a href="#B16-computers-14-00034" class="html-bibr">16</a>,<a href="#B17-computers-14-00034" class="html-bibr">17</a>].</p> Full article ">Figure 4
<p>A normal color image followed by the image modified as to emulate the seven different types of color blindness [<a href="#B18-computers-14-00034" class="html-bibr">18</a>,<a href="#B19-computers-14-00034" class="html-bibr">19</a>].</p> Full article ">Figure 5
<p>The modifications the image undertook when pre-processing with Solonko’s work [<a href="#B29-computers-14-00034" class="html-bibr">29</a>].</p> Full article ">Figure 6
<p>Original MNIST.</p> Full article ">Figure 7
<p>Fashion MNIST.</p> Full article ">Figure 8
<p>EMNIST.</p> Full article ">Figure 9
<p>The process to convert an MNIST character to an Ishihara-Like MNIST circle [<a href="#B36-computers-14-00034" class="html-bibr">36</a>].</p> Full article ">Figure 10
<p>Sample Ishihara plates from the Ishihara test. The first number inside the parenthesizes is the correct number, followed by what an individual with red-green color blindness would see [<a href="#B38-computers-14-00034" class="html-bibr">38</a>,<a href="#B39-computers-14-00034" class="html-bibr">39</a>].</p> Full article ">Figure 11
<p>A plate that only red-green color-blind individuals can decipher. The value inside the circle is 73 [<a href="#B38-computers-14-00034" class="html-bibr">38</a>].</p> Full article ">Figure 12
<p>Grayscale (<b>left</b>) and Color MNIST Ishihara Circle (<b>right</b>).</p> Full article ">Figure 13
<p>Original (<b>left</b>) and Colorized (<b>right</b>) version of an MNIST digit using the “inferno” color mask.</p> Full article ">Figure 14
<p>Program Description.</p> Full article ">Figure 15
<p>Custom 1 model in Python.</p> Full article ">Figure 16
<p>Custom 2 model in Python.</p> Full article ">Figure 17
<p>Test cases performed on the datasets.</p> Full article ">Figure 18
<p>The performance accuracy for each of the models with the grayscale Ishihara plates. The rand plate was not used in this graphic.</p> Full article ">Figure 19
<p>Confusion matrices for the Grayscale Ishihara-Like MNIST dataset showing the correct evaluation of the 10 distinct classes. Each class contained 1800 testing images in these matrices.</p> Full article ">Figure 20
<p>The performance accuracy for each of the models with each of the colored Ishihara plates. The rand plate was not used in this graphic.</p> Full article ">Figure 21
<p>Confusion matrices for the Color Ishihara-Like MNIST dataset showing the correct evaluation of the 10 distinct classes. Each class contained 1800 testing images in these matrices.</p> Full article ">Figure 22
<p>Sample images from the “random” Ishihara plate. The image on the left is 7 and the image on the right is 8.</p> Full article ">Figure 23
<p>Confusion matrices for the Color Ishihara-Like MNIST dataset without the Random plate showing the correct evaluation of the 10 distinct classes. Each class contained 1800 testing images in these matrices.</p> Full article ">Figure 24
<p>Confusion matrices for the Grayscale Ishihara-Like MNIST Random plate showing the correct evaluation of the 10 distinct classes. Each class contained 200 testing images in these matrices.</p> Full article ">Figure 25
<p>Confusion matrices for the Color Ishihara-Like MNIST Random plate showing the correct evaluation of the 10 distinct classes. Each class contained 200 testing images in these matrices.</p> Full article ">

21 pages, 3679 KiB

Open AccessArticle

Use of IoT with Deep Learning for Classification of Environment Sounds and Detection of Gases

by Priya Mishra, Naveen Mishra, Dilip Kumar Choudhary, Prakash Pareek and Manuel J. C. S. Reis

Computers 2025, 14(2), 33; https://doi.org/10.3390/computers14020033 - 22 Jan 2025

The need for safe and healthy air quality has become critical as urbanization and industrialization increase, leading to health risks and environmental concerns. Gas leaks, particularly of gases like carbon monoxide, methane, and liquefied petroleum gas (LPG), pose significant dangers due to their [...] Read more.

The need for safe and healthy air quality has become critical as urbanization and industrialization increase, leading to health risks and environmental concerns. Gas leaks, particularly of gases like carbon monoxide, methane, and liquefied petroleum gas (LPG), pose significant dangers due to their flammability and toxicity. LPG, widely used in residential and industrial settings, is especially hazardous because it is colorless, odorless, and highly flammable, making undetected leaks an explosion risk. To mitigate these dangers, modern gas detection systems employ sensors, microcontrollers, and real-time monitoring to quickly identify dangerous gas levels. This study introduces an IoT-based system designed for comprehensive environmental monitoring, with a focus on detecting LPG and butane leaks. Using sensors like the MQ6 for gas detection, MQ135 for air quality, and DHT11 for temperature and humidity, the system, managed by an Arduino Mega, collects data and sends these to the ThingSpeak platform for analysis and visualization. In cases of elevated gas levels, it triggers an alarm and notifies the user through IFTTT. Additionally, the system includes a microphone and a CNN model for analyzing audio data, enabling a thorough environmental assessment by identifying specific sounds related to ongoing activities, reaching an accuracy of 96%. Full article

(This article belongs to the Special Issue Wireless Sensor Network, IoT and Cloud Computing Technologies for Smart Cities)

► Show Figures

Figure 1

Figure 1
<p>Main block diagram of the proposed device (GAS Sensor 1 refers to MQ135 sensor and GAS Sensor 2 refers to MQ6 sensor).</p> Full article ">Figure 2
<p>Dataset of the environmental audio files, where columns are classes and rows are subclasses.</p> Full article ">Figure 3
<p>Flowchart of the ASC model.</p> Full article ">Figure 4
<p>Summary of the ASC model.</p> Full article ">Figure 5
<p>Project demonstration after collaborating IoT device with the ASC model.</p> Full article ">Figure 6
<p>Python script sending output from the Arduino to ThingSpeak channel.</p> Full article ">Figure 7
<p>Example of using ThingSpeak to visualize the environment statistics of specific days.</p> Full article ">Figure 8
<p>Example of the output prediction of the model from a random input from the testing set, in this case a rooster.</p> Full article ">Figure 9
<p>(<b>a</b>) Accuracy at the end of 30 epochs for the testing dataset (<b>b</b>) Plot representation of accuracy and loss over epochs.</p> Full article ">Figure 10
<p>Simulation showing hazardous gas detection. In these particular cases, 1 denotes “detected,” meaning that the system detected the presence of a harmful gas.</p> Full article ">

14 pages, 3057 KiB

Open AccessArticle

Leveraging Azure Automated Machine Learning and CatBoost Gradient Boosting Algorithm for Service Quality Prediction in Hospitality

by Avisek Kundu, Seeboli Ghosh Kundu, Santosh Kumar Sahu and Nitesh Dhar Badgayan

Computers 2025, 14(2), 32; https://doi.org/10.3390/computers14020032 - 22 Jan 2025

The importance of measuring service quality for business performance has been widely recognized in service marketing literature due to its pivotal influence on customer satisfaction and its long-term impact on customer loyalty. The SERVQUAL model, comprising five dimensions—reliability, assurance, tangibility, empathy, and responsiveness—provides [...] Read more.

The importance of measuring service quality for business performance has been widely recognized in service marketing literature due to its pivotal influence on customer satisfaction and its long-term impact on customer loyalty. The SERVQUAL model, comprising five dimensions—reliability, assurance, tangibility, empathy, and responsiveness—provides a measurable framework for evaluating the overall customer satisfaction. This study endeavors to ascertain whether all SERVQUAL dimensions carry equal weight in their effect on the overall service quality and to estimate the service quality based on various input features. To achieve this, questions were framed to assess the impact of variables such as gender, age, marital status, highest level of education, and frequency of hotel stays. The importance of each feature relative to the five SERVQUAL dimensions was investigated using machine learning models, specifically, CatBoost and Microsoft Azure Automated Machine Learning (AutoML) studio. This study revealed that both CatBoost and Azure AutoML identified the frequency of hotel stays and age group as the dominant predictors of service quality. Additionally, Azure AutoML highlighted the marital status as a more significant factor, suggesting its potential influence on customer preferences. The comparative modeling results demonstrated a strong alignment between the feature importance derived from CatBoost and Azure AutoML, enabling decision-makers to identify which dimensions are influenced by specific predictors and focus on targeted improvements. Full article

(This article belongs to the Special Issue AI in Its Ecosystem)

► Show Figures

Figure 1

Figure 1
<p>Architectures of (<b>a</b>) Azure AutoML [<a href="#B5-computers-14-00032" class="html-bibr">5</a>] and (<b>b</b>) CatBoost [<a href="#B6-computers-14-00032" class="html-bibr">6</a>].</p> Full article ">Figure 1 Cont.
<p>Architectures of (<b>a</b>) Azure AutoML [<a href="#B5-computers-14-00032" class="html-bibr">5</a>] and (<b>b</b>) CatBoost [<a href="#B6-computers-14-00032" class="html-bibr">6</a>].</p> Full article ">Figure 2
<p>SERVQUAL Model: factors influencing service quality and customer satisfaction [<a href="#B9-computers-14-00032" class="html-bibr">9</a>].</p> Full article ">Figure 3
<p>Workflow of AutoML in Azure AI Studio.</p> Full article ">Figure 4
<p>Decision tree model for predicting service quality based on reliability and assurance.</p> Full article ">Figure 5
<p>Correlation heat map of service quality dimensions and total score.</p> Full article ">Figure 6
<p>Feature importance results from CatBoost for service quality dimensions: (<b>a</b>) reliability, (<b>b</b>) assurance, (<b>c</b>) tangibility, (<b>d</b>) empathy, and (<b>e</b>) responsiveness.</p> Full article ">Figure 6 Cont.
<p>Feature importance results from CatBoost for service quality dimensions: (<b>a</b>) reliability, (<b>b</b>) assurance, (<b>c</b>) tangibility, (<b>d</b>) empathy, and (<b>e</b>) responsiveness.</p> Full article ">Figure 7
<p>AutoML feature importance results for service quality dimensions for (<b>a</b>) reliability, (<b>b</b>) assurance, (<b>c</b>) tangibility, (<b>d</b>) empathy, and (<b>e</b>) responsiveness.</p> Full article ">

20 pages, 2428 KiB

Open AccessArticle

Combining Smartphone Inertial Sensors and Machine Learning Algorithms to Estimate Power Variables in Standing Long Jump

by Beatrice De Lazzari, Giuseppe Vannozzi and Valentina Camomilla

Computers 2025, 14(2), 31; https://doi.org/10.3390/computers14020031 - 21 Jan 2025

Standing long jump (SLJ) power is recognized as informative of the ability of lower limbs to exert power. The study aims to provide athletes/coaches with a simple and low-cost estimate of selected SLJ power features. A group of 150 trained young participants was [...] Read more.

Standing long jump (SLJ) power is recognized as informative of the ability of lower limbs to exert power. The study aims to provide athletes/coaches with a simple and low-cost estimate of selected SLJ power features. A group of 150 trained young participants was recruited and performed a SLJ task while holding a smartphone, whose inertial sensors were used to collect data. Considering the state-of-the-art in SLJ biomechanics, a set of features was extracted and then selected by Lasso regression and used as inputs to several different optimized machine learning architectures to estimate the SLJ power variables. A Multi-Layer Perceptron Regressor was selected as the best-performing model to estimate total and concentric antero-posterior mean power, with an RMSE of 0.37 W/kg, R² > 0.70, and test phase homoscedasticity (Kendall’s τ < 0.1) in both cases. Model performance was dependent on the dataset size rather than the participants’ sex. A Multi-Layer Perceptron Regressor was able to also estimate the antero-posterior peak power (RMSE = 2.34 W/kg; R² = 0.67), although affected by heteroscedasticity. This study proved the feasibility of combining low-cost smartphone sensors and machine learning to automatically and objectively estimate SLJ power variables in ecological settings. Full article

(This article belongs to the Special Issue Applications of Machine Learning and Artificial Intelligence for Healthcare)

► Show Figures

Figure 1

Figure 1
<p>Experimental setup. Before the jump, the participant assumes the akimbo style pose over the force plate, with the SP in their right hand. The arms are maintained fixed with the hip (small black ellipse). The tape meter (big black ellipse) has the zero (highlighted with a black sign and a white arrow) near to the right heel, which corresponds to the initial position.</p> Full article ">Figure 2
<p>Workflow of the process to estimate power variables using a machine learning (ML) approach combined with SP-IMUs signals. The flowchart represents the steps from extraction of signals (SP-IMUs and force plate, <a href="#sec2dot1-computers-14-00031" class="html-sec">Section 2.1</a>) to machine learning models (ML models, <a href="#sec2dot4-computers-14-00031" class="html-sec">Section 2.4</a>), with analysis of the optimized models (<a href="#sec2dot5-computers-14-00031" class="html-sec">Section 2.5</a> and <a href="#sec2dot6-computers-14-00031" class="html-sec">Section 2.6</a>).</p> Full article ">Figure 3
<p>Ground Reaction Force (GRF) derived: (<b>a</b>) vertical component of the acceleration (a<sub>CoM,V</sub>) and power (P<sub>V</sub>); (<b>b</b>) antero-posterior component acceleration (a<sub>CoM,AP</sub>) and power (P<sub>AP</sub>). Dashed vertical lines define the concentric (from t<sub>0</sub> to t<sub>BP</sub>) and eccentric phases (from t<sub>BP</sub> to t<sub>TO</sub>).</p> Full article ">Figure 4
<p>Vertical acceleration (a<sub>V</sub>) and velocity (v<sub>V</sub>) extracted from SP-IMUs with highlighted phases: unloading (t<sub>0</sub> – t<sub>UL</sub>), eccentric yielding (t<sub>UL</sub> – t<sub>UB</sub>), eccentric braking (t<sub>UL</sub> – t<sub>BP</sub>), and concentric propulsive phase (t<sub>BP</sub> – t<sub>TO</sub>). Legend for time instants: t<sub>0</sub> = jump onset; t<sub>UL</sub> = minimum acceleration; t<sub>UB</sub> = minimum velocity; t<sub>BP</sub> = velocity crosses 0; t<sub>TO</sub> = take-off.</p> Full article ">Figure 5
<p>Bland-Altman plots on the test set: (<b>a</b>) P<sub>AP, tot</sub>; (<b>b</b>) P<sub>AP, conc</sub>. Legend: upper limit (UL = BIAS + 1.96 × SD), lower limit (LL == BIAS − 1.96 × SD), confidence intervals (CI) at 95% of BIAS, UL, and LL were calculated following [<a href="#B43-computers-14-00031" class="html-bibr">43</a>], with BIAS = test value—model predicted value; SD = standard deviation of the previous differences. In both plots, the equation and R<sup>2</sup> show values lower than 0.3.</p> Full article ">Figure 6
<p>PFI performed on MLPR model used to estimate P<sub>AP, tot</sub>. In the legend, the 32 features used to estimate P<sub>AP, tot</sub> divided into four groups in terms of PFI output. The first two groups (dark grey) showed an MSE<sub>i</sub>/MSE<sub>0</sub> greater than 1, meaning that the features belonging to those groups affect the estimate of the output more. The third group (light grey) showed a MSE<sub>i</sub>/MSE<sub>0</sub> equal to 1, while the fourth group (black) showed a MSE<sub>i</sub>/MSE<sub>0</sub> lower than 1.</p> Full article ">Figure 7
<p>PFI performed on MLPR model used to estimate P<sub>AP conc</sub>. In the legend, the 48 features used to estimate P<sub>AP, conc</sub> divided into seven groups in terms of PFI output. The first three groups (dark grey) showed an MSE<sub>i</sub>/MSE<sub>0</sub> greater than 1, meaning that the features belonging to those groups affect the estimate of the output more. The fourth group (light grey) showed a MSE<sub>i</sub>/MSE<sub>0</sub> equal to 1, while the black groups showed a MSE<sub>i</sub>/MSE<sub>0</sub> lower than 1.</p> Full article ">

35 pages, 633 KiB

Open AccessArticle

Set-Word Embeddings and Semantic Indices: A New Contextual Model for Empirical Language Analysis

by Pedro Fernández de Córdoba, Carlos A. Reyes Pérez, Claudia Sánchez Arnau and Enrique A. Sánchez Pérez

Computers 2025, 14(1), 30; https://doi.org/10.3390/computers14010030 - 20 Jan 2025

We present a new word embedding technique in a (non-linear) metric space based on the shared membership of terms in a corpus of textual documents, where the metric is naturally defined by the Boolean algebra of all subsets of the corpus and a [...] Read more.

We present a new word embedding technique in a (non-linear) metric space based on the shared membership of terms in a corpus of textual documents, where the metric is naturally defined by the Boolean algebra of all subsets of the corpus and a measure

μ

defined on it. Once the metric space is constructed, a new term (a noun, an adjective, a classification term) can be introduced into the model and analyzed by means of semantic projections, which in turn are defined as indexes using the measure

μ

and the word embedding tools. We formally define all necessary elements and prove the main results about the model, including a compatibility theorem for estimating the representability of semantically meaningful external terms in the model (which are written as real Lipschitz functions in the metric space), proving the relation between the semantic index and the metric of the space (Theorem 1). Our main result proves the universality of our word-set embedding, proving mathematically that every word embedding based on linear space can be written as a word-set embedding (Theorem 2). Since we adopt an empirical point of view for the semantic issues, we also provide the keys for the interpretation of the results using probabilistic arguments (to facilitate the subsequent integration of the model into Bayesian frameworks for the construction of inductive tools), as well as in fuzzy set-theoretic terms. We also show some illustrative examples, including a complete computational case using big-data-based computations. Thus, the main advantages of the proposed model are that the results on distances between terms are interpretable in semantic terms once the semantic index used is fixed and, although the calculations could be costly, it is possible to calculate the value of the distance between two terms without the need to calculate the whole distance matrix. “Wovon man nicht sprechen kann, darüber muss man schweigen”. Tractatus Logico-Philosophicus. L. Wittgenstein. Full article

(This article belongs to the Special Issue Natural Language Processing (NLP) and Large Language Modelling)

► Show Figures

Figure 1

Figure 1
<p>Word embedding versus set-word embedding.</p> Full article ">Figure 2
<p>Sequences <span class="html-italic">s</span> in a text for <math display="inline"><semantics> <mrow> <mi>n</mi> <mo>=</mo> <mn>6</mn> <mo>.</mo> </mrow> </semantics></math> The last circle marks the sequence <math display="inline"><semantics> <msub> <mi>s</mi> <mn>0</mn> </msub> </semantics></math>, which contains the set <math display="inline"><semantics> <mrow> <mi>H</mi> <mo>(</mo> <msub> <mi>s</mi> <mn>0</mn> </msub> <mo>)</mo> <mo>=</mo> </mrow> </semantics></math> {“<span class="html-italic">coin</span>”, “<span class="html-italic">and</span>”, “<span class="html-italic">a</span>”, “<span class="html-italic">silver</span>”, “<span class="html-italic">that</span>”}, and the previous one, <math display="inline"><semantics> <msub> <mi>s</mi> <mn>1</mn> </msub> </semantics></math>, contains the words <math display="inline"><semantics> <mrow> <mi>H</mi> <mo>(</mo> <msub> <mi>s</mi> <mn>1</mn> </msub> <mo>)</mo> <mo>=</mo> </mrow> </semantics></math> {“<span class="html-italic">gold</span>”, “<span class="html-italic">coin</span>”, “<span class="html-italic">and</span>”, “<span class="html-italic">a</span>”, “<span class="html-italic">silver</span>”}.</p> Full article ">Figure 3
<p>Sets defining the representation of the set-word embedding.</p> Full article ">Figure 4
<p>A 3D representation of the embedding of the closest words to the term “gold”, according to the Euclidean distance, using Word2vec Google News (71291x200) (<a href="http://projector.tensorflow.org/" target="_blank">http://projector.tensorflow.org/</a> (accessed on 3 January 2025)).</p> Full article ">Figure 5
<p>Semantic indices <math display="inline"><semantics> <msub> <mi>P</mi> <mi>A</mi> </msub> </semantics></math> for <math display="inline"><semantics> <mrow> <mi>A</mi> <mo>=</mo> </mrow> </semantics></math> gold (<math display="inline"><semantics> <mrow> <mi>q</mi> <mo>=</mo> <msub> <mi>q</mi> <mi>μ</mi> </msub> </mrow> </semantics></math>).</p> Full article ">Figure 6
<p>Conjugate semantic indices <math display="inline"><semantics> <msub> <mi>P</mi> <mi>B</mi> </msub> </semantics></math> for <span class="html-italic">B</span> for each of the selected words.</p> Full article ">Figure 7
<p>Comparison of the results provided by Word2vec and the set-word embedding defined by Google search for the term “gold”.</p> Full article ">Figure 8
<p>Comparison of the results provided by Word2vec and the set-word embedding excluding the term “blue” for a better visualization.</p> Full article ">Figure 9
<p>Comparison of the results provided by Word2vec and the set-word embedding including the other word embeddings of <a href="#computers-14-00030-t002" class="html-table">Table 2</a>.</p> Full article ">Figure 10
<p>Comparison of the results provided by the set-word embedding and the models E5-small and MiniLM-L6, which are the ones that have a similar distribution.</p> Full article ">

33 pages, 19016 KiB

Open AccessArticle

Multitask Learning-Based Pipeline-Parallel Computation Offloading Architecture for Deep Face Analysis

by Faris S. Alghareb and Balqees Talal Hasan

Computers 2025, 14(1), 29; https://doi.org/10.3390/computers14010029 - 20 Jan 2025

Deep Neural Networks (DNNs) have been widely adopted in several advanced artificial intelligence applications due to their competitive accuracy to the human brain. Nevertheless, the superior accuracy of a DNN is achieved at the expense of intensive computations and storage complexity, requiring custom [...] Read more.

Deep Neural Networks (DNNs) have been widely adopted in several advanced artificial intelligence applications due to their competitive accuracy to the human brain. Nevertheless, the superior accuracy of a DNN is achieved at the expense of intensive computations and storage complexity, requiring custom expandable hardware, i.e., graphics processing units (GPUs). Interestingly, leveraging the synergy of parallelism and edge computing can significantly improve CPU-based hardware platforms. Therefore, this manuscript explores levels of parallelism techniques along with edge computation offloading to develop an innovative hardware platform that improves the efficacy of deep learning computing architectures. Furthermore, the multitask learning (MTL) approach is employed to construct a parallel multi-task classification network. These tasks include face detection and recognition, age estimation, gender recognition, smile detection, and hair color and style classification. Additionally, both pipeline and parallel processing techniques are utilized to expedite complicated computations, boosting the overall performance of the presented deep face analysis architecture. A computation offloading approach, on the other hand, is leveraged to distribute computation-intensive tasks to the server edge, whereas lightweight computations are offloaded to edge devices, i.e., Raspberry Pi 4. To train the proposed deep face analysis network architecture, two custom datasets (HDDB and FRAED) were created for head detection and face-age recognition. Extensive experimental results demonstrate the efficacy of the proposed pipeline-parallel architecture in terms of execution time. It requires 8.2 s to provide detailed face detection and analysis for an individual and 23.59 s for an inference containing 10 individuals. Moreover, a speedup of 62.48% is achieved compared to the sequential-based edge computing architecture. Meanwhile, 25.96% speed performance acceleration is realized when implementing the proposed pipeline-parallel architecture only on the server edge compared to the sever sequential implementation. Considering classification efficiency, the proposed classification modules achieve an accuracy of 88.55% for hair color and style classification and a remarkable prediction outcome of 100% for face recognition and age estimation. To summarize, the proposed approach can assist in reducing the required execution time and memory capacity by processing all facial tasks simultaneously on a single deep neural network rather than building a CNN model for each task. Therefore, the presented pipeline-parallel architecture can be a cost-effective framework for real-time computer vision applications implemented on resource-limited devices. Full article

► Show Figures

Figure 1

Figure 1
<p>Head detection dataset versus face detection dataset using nano-based YOLOv8.</p> Full article ">Figure 2
<p>Sample images from the hair dataset used to train the hair color-style module.</p> Full article ">Figure 3
<p>Selected image samples of the created face recognition and age estimation dataset.</p> Full article ">Figure 4
<p>The general framework of the proposed deep face analysis architecture.</p> Full article ">Figure 5
<p>Stages of the pipeline-multithreading architecture, showing four images being processed in parallel.</p> Full article ">Figure 6
<p>Proposed pipeline-parallel architectures with thread distributions; (<b>a</b>) multithreading three MTL-based classifiers on a single edge device, (<b>b</b>) multithreading three MTL-based classifiers on a cluster containing three edge computing devices.</p> Full article ">Figure 7
<p>Modified VGG-Face network to support the multitask classification approach.</p> Full article ">Figure 8
<p>Offloading feature maps of detected heads to edge devices using multithreading.</p> Full article ">Figure 9
<p>Multithreading of parallel modules on edge server and edge node processors.</p> Full article ">Figure 10
<p>The framework of system deployment for the proposed deep face analysis.</p> Full article ">Figure 11
<p>Training and validation performance of the YOLOv8 model for head detection. The x-axis represents the number of epochs.</p> Full article ">Figure 12
<p>YOLOv8 testing performance; (<b>a</b>) confusion matrix, (<b>b</b>) precision, (<b>c</b>) recall, (<b>d</b>) precision-recall, and (<b>e</b>) F1 score confidence curve.</p> Full article ">Figure 13
<p>Head detection result samples of YOLOv8, where a red box denotes a detected head with its corresponding confidence level.</p> Full article ">Figure 14
<p>Confusion matrices for classification modules using STL and MTL; (<b>a</b>) hair color STL, (<b>b</b>) hair color MTL, (<b>c</b>) hairstyle STL, (<b>d</b>) hairstyle MTL, (<b>e</b>) gender STL, (<b>f</b>) gender MTL, (<b>g</b>) smile STL, (<b>h</b>) smile MTL, (<b>i</b>) Face STL, (<b>j</b>) Face MTL, (<b>k</b>) age STL, and (<b>l</b>) age MTL module.</p> Full article ">Figure 15
<p>Speed performance evaluation of the proposed pipeline-parallel architecture; (<b>a</b>) execution time for pipeline-parallel configurations versus sequential implementation, (<b>b</b>) speedup comparisons of implemented configurations.</p> Full article ">

18 pages, 731 KiB

Open AccessReview

Computational Methods for Information Processing from Natural Language Complaint Processes—A Systematic Review

by J. C. Blandón Andrade, A. Castaño Toro, A. Morales Ríos and D. Orozco Ospina

Computers 2025, 14(1), 28; https://doi.org/10.3390/computers14010028 - 20 Jan 2025

Complaint processing is of great importance for companies because it allows them to understand customer satisfaction levels, which is crucial for business success. It allows them to show the real perceptions of users and thus visualize the problems, which are regularly processed from [...] Read more.

Complaint processing is of great importance for companies because it allows them to understand customer satisfaction levels, which is crucial for business success. It allows them to show the real perceptions of users and thus visualize the problems, which are regularly processed from oral or written natural language, derived from the provision of a service. In addition, the treatment of complaints is relevant because according to the laws of each country, companies have the obligation to respond to these complaints in a specified time. The specialized literature mentions that enterprises lost USD 75 billion due to poor customer service, highlighting that companies need to know and understand customer perceptions, especially emotions, and product reviews to gain insight and learn about customer feedback because of the importance of the voice of the customer for an organization. In general, it is evident that there is a need for research related to computational language processing to handle user requests. The authors show great interest in computational techniques for the processing of this information in natural language and how this could contribute to the improvement of processes within the productive sector. This work searches in indexed journals for information related to computational methods for processing relevant data from user complaints. It is proposed to apply a systematic literature review (SLR) method combining literature review guides by Kitchenham and the PRISMA statement. The systematic process allows the extraction of consistent information, and after applying it, 27 articles were obtained from which the analysis was conducted. The results show various proposals using linguistic, statistical, machine learning, and hybrid methods. We find that most authors combine Natural Language Processing (NLP) and Machine Learning (ML) to create hybrid methods. The methods extract relevant information from complaints of the customers in natural language in various domains, such as government, medical, banks, e-commerce, public services, agriculture, customer service, environmental, and tourism, among others. This work contributes as support for the creation of new systems that can give companies a significant competitive advantage due to their ability to reduce the response time of the complaints as established by law. Full article

► Show Figures

Figure 1

40 pages, 21233 KiB

Open AccessArticle

Large-Scale Cross-Cultural Tourism Analytics: Integrating Transformer-Based Text Mining and Network Analysis

by Dian Puteri Ramadhani, Andry Alamsyah, Mochamad Yudha Febrianta, Muhammad Nadhif Fajriananda, Mahira Shafiya Nada and Fathiyyah Hasanah

Computers 2025, 14(1), 27; https://doi.org/10.3390/computers14010027 - 16 Jan 2025

The growth of the tourism industry in Southeast Asia, particularly in Indonesia, Thailand, and Vietnam, establishes the region as a leading global tourism destination. Numerous studies have explored tourist behavior within specific regions. However, the question of whether tourists’ experience perceptions differ based [...] Read more.

The growth of the tourism industry in Southeast Asia, particularly in Indonesia, Thailand, and Vietnam, establishes the region as a leading global tourism destination. Numerous studies have explored tourist behavior within specific regions. However, the question of whether tourists’ experience perceptions differ based on their cultural backgrounds is still insufficiently addressed. Previous articles suggest that an individual’s cultural background plays a significant role in shaping tourist values and expectations. This study investigates how tourists’ cultural backgrounds, represented by their geographical regions of origin, impact their entertainment experiences, sentiments, and mobility patterns across the three countries. We gathered 387,010 TripAdvisor reviews and analyzed them using a combination of advanced text mining techniques and network analysis to map tourist mobility patterns. Comparing sentiments and behaviors across cultural backgrounds, this study found that entertainment preferences vary by origin. The network analysis reveals distinct exploration patterns: diverse and targeted exploration. Vietnam achieves the highest satisfaction across the cultural groups through balanced development, while Thailand’s integrated entertainment creates cultural divides, and Indonesia’s generates moderate satisfaction regardless of cultural background. This study contributes to understanding tourism dynamics in Southeast Asia through a data-driven, comparative analysis of tourist behaviors. The findings provide insights for destination management, marketing strategies, and policy development, highlighting the importance of tailoring tourism offerings to meet the diverse preferences of visitors from different global regions. Full article

(This article belongs to the Special Issue Recent Advances in Social Networks and Social Media)

► Show Figures

Figure 1

Figure 1
<p>Research workflow.</p> Full article ">Figure 2
<p>Number of tourist reviews for Indonesia, Thailand, and Vietnam.</p> Full article ">Figure 3
<p>Proportions for tourists’ non-entertainment (Non-E) and entertainment experiences, including nightlife and festive entertainment (NFE), recreational entertainment (RCE), nature-based entertainment (NBE), culinary entertainment (CNE), and cultural entertainment (CTE) for (<b>a</b>) Indonesia, (<b>b</b>) Thailand, and (<b>c</b>) Vietnam.</p> Full article ">Figure 4
<p>Cross-regional analysis of tourist sentiment distribution.</p> Full article ">Figure 5
<p>Comparative sentiment distribution across entertainment dimensions in Indonesia, Thailand, and Vietnam. “x” symbols indicate the mean values for each category while dots represent outlier points that fall outside the normal distribution range.</p> Full article ">Figure 6
<p>Tourist sentiment comparison across entertainment categories in Indonesia, Thailand, and Vietnam, by origin.</p> Full article ">Figure 7
<p>Tourist mobility in Indonesia by origin. Tourist destinations in Indonesia are abbreviated as follows: Waterbom Bali (WTB), Sacred Monkey Forest Sanctuary (SMF), Tegalalang Rice Terrace (TRT), Tanah Lot Temple (TLT), Bali Zoo (BZO), Uluwatu Temple (ULT), Borobudur Temple (BRT), Prambanan Temples (PRT), Ijen Crater (ICR), Campuhan Ridge Walk (CRW), Tirta Empul Temple (TET), Mount Batur (MBA), Nusa Dua Beach (NDB), Mount Bromo (MBR), Tirta Gangga (TGG), Bali Safari & Marine Park (BSM), Sanur Beach (SNB), Kelingking Beach (KLB), Bali Bird Park (BBP), Jatiluwih Green Land (JGL), Ulun Danu Bratan Temple (UDB), National Monument/MONAS (MON), Museum PASIFIKA (MPS), Kuta Beach (KTB), and Seminyak Beach (SMB). Colors indicate destination categories: cultural sites (green), natural attractions (purple), and urban recreational spaces (orange).</p> Full article ">Figure 8
<p>Tourist mobility in Thailand by origin. Tourist destinations in Thailand are abbreviated as follows: Wat Phra Chetuphon (WPC), BTS Skytrain (BTS), The Grand Palace (TGP), Chatuchak Weekend Market (CWM), Temple of Dawn/Wat Arun (TOD), Big Buddha Phuket (BBP), Temple of the Emerald Buddha (TEB), Jim Thompson House (JTH), Siam Paragon (SP), The Sanctuary of Truth (TST), Bangla Road (BR), Green Elephant Sanctuary Park (GEP), Wat Chedi Luang Varavihara (WLV), Temple of the Golden Buddha (TGB), Khaosan Road (KR), Wat Phra That Doi Suthep (WTS), Wat Rong Khun (WRK), Lumpini Park (LP), Kata Beach (KB), Banana Beach (BB), Patong Beach (PB), Bridge Over the River Kwai (BRK), SEA LIFE Bangkok Ocean World (SL), Safari World (SW), and Tiger Kingdom (TK). Colors indicate destination categories: cultural sites (green), natural attractions (purple), and urban recreational spaces (orange).</p> Full article ">Figure 9
<p>Tourist mobility in Vietnam by origin. Tourist destinations in Vietnam are abbreviated as follows: Old Quarter (OQ), War Remnants Museum (WRM), Cu Chi Tunnels (CCT), Hoi An Ancient Town (HAT), Halong Bay (HB), Lake of the Restored Sword (HKL), Hue Imperial City (HIC), Temple of Literature & National University (TLU), The Marble Mountains (TMM), Hoa Lo Prison (HLP), Central Post Office (CPO), Ho Chi Minh Mausoleum (HCM), An Bang Beach (ABB), Vietnamese Women’s Museum (VWM), Vietnam Museum of Ethnology (VME), The Independence Palace (TIP), Lady Buddha (LB), Dragon Bridge (DB), Po Nagar Cham Towers (PNC), Bitexco Financial Tower (BFT), My Son Sanctuary (MSS), Japanese Covered Bridge (JCB), Saigon Notre Dame Cathedral (SND), Thien Mu Pagoda (TMP), and Tam Coc (TC). Colors indicate destination categories: cultural sites (green), natural attractions (purple), and urban recreational spaces (orange).</p> Full article ">Figure 10
<p>Comparison of network entropy.</p> Full article ">Figure 11
<p>Comparison of network efficiency.</p> Full article ">

17 pages, 7797 KiB

Open AccessArticle

A 32 μm² MOS-Based Remote Sensing Temperature Sensor with 1.29 °C Inaccuracy for Thermal Management

by Ruohan Yang, Kwabena Oppong Banahene, Bryce Gadogbe, Randall Geiger and Degang Chen

Computers 2025, 14(1), 26; https://doi.org/10.3390/computers14010026 - 16 Jan 2025

This paper introduces a compact NMOS-based temperature sensor designed for precise thermal management in high-performance integrated circuits. Fabricated using the TSMC 180 nm process with a 1.8 V supply, this sensor employs a single diode-connected NMOS transistor, achieving a significant size reduction and [...] Read more.

This paper introduces a compact NMOS-based temperature sensor designed for precise thermal management in high-performance integrated circuits. Fabricated using the TSMC 180 nm process with a 1.8 V supply, this sensor employs a single diode-connected NMOS transistor, achieving a significant size reduction and improved voltage headroom. The sensor’s area is 32 µm² per unit, enabling dense integration around thermal hotspots. A novel voltage calibration method ensures accurate temperature extraction. The measurement results demonstrate three-sigma errors within ±0.1 °C in the critical range of 75 °C to 95 °C and +1.29/−1.08 °C outside this range, confirming the sensor’s high accuracy and suitability for advanced thermal management applications. Full article

► Show Figures

Figure 1

Figure 1
<p>Thermal masks for temperature sensors of JEDEC 21-C Grade B Standard [<a href="#B22-computers-14-00026" class="html-bibr">22</a>] and the target mask for this work.</p> Full article ">Figure 2
<p>Temperature measurement system with distributed sensor units across multiple cores.</p> Full article ">Figure 3
<p>The schematic of the temperature sensing circuit with the Banba bandgap reference circuit as the Temperature Measurement Controller (TMC) circuit.</p> Full article ">Figure 4
<p>The schematic of the temperature sensor unit.</p> Full article ">Figure 5
<p>Photograph of the chip of the temperature measurement system (highlighted within the red dashed block).</p> Full article ">Figure 6
<p>Layout of the temperature measurement system with distances between 36 sensor units.</p> Full article ">Figure 7
<p>Test instrument setup.</p> Full article ">Figure 8
<p>The temperature errors of 180 sensors on 5 DUTs with external instrument current sources after individual calibration at 80 °C and 90 °C.</p> Full article ">Figure 9
<p>Relationship between TMC currents and temperatures.</p> Full article ">Figure 10
<p>The temperature errors of 36 sensors on 1 DUT with the internal current source TMC on the same chip after individual calibration at 80 °C and 90 °C.</p> Full article ">Figure 11
<p>The temperature errors of 180 sensors on 5 DUTs with temperature differences between the TMC and sensors after individual calibration at 80 °C and 90 °C.</p> Full article ">Figure 12
<p>The temperature errors in three cases after individual calibration at 80 °C and 90 °C within the critical range of 75 °C to 95 °C.</p> Full article ">Figure 13
<p>The temperature errors in three cases after individual calibration at 80 °C and 90 °C across the wider temperature range of −10 °C to 100 °C.</p> Full article ">

18 pages, 4133 KiB

Open AccessArticle

An Investigation of Hand Gestures for Controlling Video Games in a Rehabilitation Exergame System

by Radhiatul Husna, Komang Candra Brata, Irin Tri Anggraini, Nobuo Funabiki, Alfiandi Aulia Rahmadani and Chih-Peng Fan

Computers 2025, 14(1), 25; https://doi.org/10.3390/computers14010025 - 15 Jan 2025

Musculoskeletal disorders (MSDs) can significantly impact individuals’ quality of life (QoL), often requiring effective rehabilitation strategies to promote recovery. However, traditional rehabilitation methods can be expensive and may lack engagement, leading to poor adherence to therapy exercise routines. An exergame system can [...] Read more.

Musculoskeletal disorders (MSDs) can significantly impact individuals’ quality of life (QoL), often requiring effective rehabilitation strategies to promote recovery. However, traditional rehabilitation methods can be expensive and may lack engagement, leading to poor adherence to therapy exercise routines. An exergame system can be a solution to this problem. In this paper, we investigate appropriate hand gestures for controlling video games in a rehabilitation exergame system. The Mediapipe Python library is adopted for the real-time recognition of gestures. We choose 10 easy gestures among 32 possible simple gestures. Then, we specify and compare the best and the second-best groups used to control the game. Comprehensive experiments are conducted with 16 students at Andalas University, Indonesia, to find appropriate gestures and evaluate user experiences of the system using the System Usability Scale (SUS) and User Experience Questionnaire (UEQ). The results show that the hand gestures in the best group are more accessible than in the second-best group. The results suggest appropriate hand gestures for game controls and confirm the proposal’s validity. In future work, we plan to enhance the exergame system by integrating a diverse set of video games, while expanding its application to a broader and more diverse sample. We will also study other practical applications of the hand gesture control function. Full article

► Show Figures

Figure 1

25 pages, 1993 KiB

Open AccessArticle

Hacking Exposed: Leveraging Google Dorks, Shodan, and Censys for Cyber Attacks and the Defense Against Them

by Abdullah Alabdulatif and Navod Neranjan Thilakarathne

Computers 2025, 14(1), 24; https://doi.org/10.3390/computers14010024 - 15 Jan 2025

In recent years, cyberattacks have increased in sophistication, using a variety of tools to exploit vulnerabilities across the global digital landscapes. Among the most commonly used tools at an attacker’s disposal are Google dorks, Shodan, and Censys, which offer unprecedented access to exposed [...] Read more.

In recent years, cyberattacks have increased in sophistication, using a variety of tools to exploit vulnerabilities across the global digital landscapes. Among the most commonly used tools at an attacker’s disposal are Google dorks, Shodan, and Censys, which offer unprecedented access to exposed systems, devices, and sensitive data on the World Wide Web. While these tools can be leveraged by professional hackers, they have also empowered “Script Kiddies”, who are low-skill, inexperienced attackers who use readily available exploits and scanning tools without deep technical knowledge. Consequently, cyberattacks targeting critical infrastructure are growing at a rapid rate, driven by the ease with which these solutions can be operated with minimal expertise. This paper explores the potential for cyberattacks enabled by these tools, presenting use cases where these platforms have been used for both offensive and defensive purposes. By examining notable incidents and analyzing potential threats, we outline proactive measures to protect against these emerging risks. In this study, we delve into how these tools have been used offensively by attackers and how they serve defensive functions within cybersecurity. Additionally, we also introduce an automated all-in-one tool designed to consolidate the functionalities of Google dorks, Shodan, and Censys, offering a streamlined solution for vulnerability detection and analysis. Lastly, we propose proactive defense strategies to mitigate exploitation risks associated with such tools, aiming to enhance the resilience of critical digital infrastructure against evolving cyber threats. Full article

(This article belongs to the Special Issue Multimedia Data and Network Security)

► Show Figures

Figure 1

27 pages, 1409 KiB

Open AccessArticle

Adaptive Handover Management in High-Mobility Networks for Smart Cities

by Yahya S. Junejo, Faisal K. Shaikh, Bhawani S. Chowdhry and Waleed Ejaz

Computers 2025, 14(1), 23; https://doi.org/10.3390/computers14010023 - 14 Jan 2025

The seamless handover of mobile devices is critical for maximizing the potential of smart city applications, which demand uninterrupted connectivity, ultra-low latency, and performance in diverse environments. Fifth-generation (5G) and beyond-5G networks offer advancements in massive connectivity and ultra-low latency by leveraging advanced [...] Read more.

The seamless handover of mobile devices is critical for maximizing the potential of smart city applications, which demand uninterrupted connectivity, ultra-low latency, and performance in diverse environments. Fifth-generation (5G) and beyond-5G networks offer advancements in massive connectivity and ultra-low latency by leveraging advanced technologies like millimeter wave, massive machine-type communication, non-orthogonal multiple access, and beam forming. However, challenges persist in ensuring smooth handovers in dense deployments, especially in higher frequency bands and with increased user mobility. This paper presents an adaptive handover management scheme that utilizes reinforcement learning to optimize handover decisions in dynamic environments. The system selects the best target cell from the available neighbor cell list by predicting key performance indicators, such as reference signal received power and the signal–interference–noise ratio, while considering the fixed time-to-trigger and hysteresis margin values. It dynamically adjusts handover thresholds by incorporating an offset based on real-time network conditions and user mobility patterns. This adaptive approach minimizes handover failures and the ping-pong effect. Compared to the baseline LIM2 model, the proposed system demonstrates a 15% improvement in handover success rate, a 3% improvement in user throughput, and an approximately 6 sec reduction in the latency at 200 km/h speed in high-mobility scenarios. Full article

(This article belongs to the Special Issue Wireless Sensor Network, IoT and Cloud Computing Technologies for Smart Cities)

► Show Figures

Figure 1

Figure 1
<p>Learning model for handover in 5G NR.</p> Full article ">Figure 2
<p>Handover signaling messages.</p> Full article ">Figure 3
<p>Proposed network mechanism.</p> Full article ">Figure 4
<p>Handover event A5.</p> Full article ">Figure 5
<p>Handover success rate when total number of UE is 500 versus (<b>a</b>) speed and (<b>b</b>) SINR.</p> Full article ">Figure 6
<p>UE handover delay versus (<b>a</b>) speed and (<b>b</b>) SINR.</p> Full article ">Figure 7
<p>Throughput versus (<b>a</b>) speed, (<b>b</b>) SINR, and (<b>c</b>) when number of pieces of UE is fixed.</p> Full article ">Figure 8
<p>Latency versus (<b>a</b>) speed and (<b>b</b>) SINR.</p> Full article ">Figure 9
<p>Packet loss ratio versus (<b>a</b>) speed and (<b>b</b>) SINR.</p> Full article ">

More Articles...

Submit to Computers Review for Computers

Journal Menu

Journal Browser

► Journal Browser

Highly Accessed Articles

View More...

Latest Books

More Books and Reprints...

E-Mail Alert

News

24 January 2025
Meet Us at the 30^th Annual ACM Conference on Intelligent User Interfaces, 24–27 March 2025, Cagliari, Italy

21 January 2025
MDPI’s Newly Launched Journals in December 2024

7 January 2025
MDPI INSIGHTS: The CEO's Letter #19 - Reflecting on 2024, Society Journals, OA Germany

More News & Announcements...

Topics

Propose a Topic

Topic in Computers, Energies, Future Internet, Information, Mathematics

Research on Blockchain Technology for Peer-to-Peer (P2P) Energy Trading Topic Editors: Pierre-Martin Tardif, Brahim El Bhiri, Bilal Abu-Salih, Kenji Tanaka
Deadline: 28 February 2025

Topic in Applied Sciences, Computers, Electronics, Information, J. Imaging

Visual Computing and Understanding: New Developments and Trends Topic Editors: Wei Zhou, Guanghui Yue, Wenhan Yang
Deadline: 30 March 2025

Topic in AI, Buildings, Computers, Drones, Entropy, Symmetry

Applications of Machine Learning in Large-Scale Optimization and High-Dimensional Learning Topic Editors: Jeng-Shyang Pan, Junzo Watada, Vaclav Snasel, Pei Hu
Deadline: 30 April 2025

Topic in Applied Sciences, ASI, Blockchains, Computers, MAKE, Software

Recent Advances in AI-Enhanced Software Engineering and Web Services Topic Editors: Hai Wang, Zhe Hou
Deadline: 31 May 2025

More Topics

Conferences

Announce Your Conference

16–19 May 2025 The 5th International Conference on Symmetry

29 July–3 August 2025 The 2nd International Conference on AI Sensors and Transducers

More Conferences...

Special Issues

Propose a Special Issue

Special Issue in Computers

Uncertainty-Aware Artificial Intelligence Guest Editors: Hussain Mohammed Dipu Kabir, Syed Bahauddin Alam, Subrota Kumar Mondal, Jeremy Straub
Deadline: 31 January 2025

Special Issue in Computers

Middleware for the Computing Continuum 2024 Guest Editors: Giuseppe Di Modica, Paolo Bellavista, Eirini Eleni Tsiropoulou, Antonino Galletta
Deadline: 31 January 2025

Special Issue in Computers

AI in Its Ecosystem Guest Editors: Behrouz Zolfaghari, Amin Ramezani, Mohsen Hadian, Firooz Saghezchi
Deadline: 1 February 2025

Special Issue in Computers

Cloud Computing and Big Data Mining Guest Editor: Rao Mikkilineni
Deadline: 28 February 2025

More Special Issues

Back to TopTop