Dharmender Salian, Department of Information Technology, University of the Cumberlands, New York, USA
This study aims to review studies related to Artificial Intelligence (AI) maturity models (MM) in automotive manufacturing systematically. Digitalization is required for automotive small and medium enterprises (SMEs) if they want to remain competitive. In the USA, a large segment of workers is employed in SMEs. Due to scarce funds, SMEs had not been aggressive in digitalization but benefits like operational efficiency, quality improvement, cost reduction and innovative culture make it appealing and beneficial to customers. Digitalization of operations that employ Artificial Intelligence techniques is becoming more commonplace in the past few years. Using the lens of an AI maturity model, this paper reviews the state of AI applications in SMEs.
Industry 5.0, Maturity Model, AI maturity model, Maturity level & digitalization.
Aicha Aggoune, Department of Computer Science, LabSTIC, University of 8th May 1945, Guelma, Algeria
The NoSQL databases allow the storage and processing of huge data in a distributed environment, ensuring high availability, scalability, and fault tolerance. Unlike relational databases, where the data are presented by the relational model, the NoSQL databases provide four data models that are different from one another (key-value, Columnar, Document, and Graph). In this context, novel opportunities may arise when leveraging the selection of the NoSQL database, which may be more suitable than another for representing such data. This paper presents the MCMM algorithm for transforming the MongoDB database to the Cassandra database and vice versa. The proposed algorithm is based on the use of translation and inverse translation rules. Two examples are used to demonstrate how the MCMM algorithm works.
MongoDB, Cassandra, Translation rules, Inverse translation rules, Mapping between NoSQL.
Hiro Horie1, Sinan Chen1, Masahide Nakamura1,2, and Kiyoshi Yasuda3, 1Kobe University, 1-1 Rokkodai-cho, Nada, Kobe, 657-8501, Japan, 2RIKEN Center for Advanced Intelligence Project, 1-4-1 Nihonbashi, Chuo-ku, Tokyo, 103-0027, Japan, 3Osaka Institute of Technology 5-16-1 Omiya, Asahi-ku, Osaka, 535–8585 Japan
Our research group is conducting research of a system to support the lives of the elderly at home. We developed “Rakuraku Video Service”(Rakuraku means easy.), a service that obtains information on the interests and preferences of elderly people and recommends YouTube videos based on this information. The purpose of this service is to help the elderly relieve stress and relax by watching videos. However, it has not been tested yet whether this service has such an effect on the elderly. We conduct an experiment to evaluate the service. In conducting the experiment, we will collaborate with a previous study, “PC-Mei” which aims to watch over the elderly and support their daily lives with a virtual agent. The experiment was conducted to obtain evaluations of watched videos and a questionnaire. From the results, it is clear that the service is useful in relieving stress among the elderly.
Elderly at home, watching videos, stress relief, individual adaptive type, spoken dialogue agent.
Aoi Nagatani1, Tasuku Watanabe1, Yuya Tarutani1, Yoshifumi Kamae1, Shun Sato1, Marin Shoda1, and Masahide Nakamura2, 1Graduate School of System Informatics, Kobe University Rokkodai-cho 1–1，Nada-ku，Kobe，Hyogo，657–8501 Japan, 2the Center of Mathematical and Data Science, Kobe Univ.
In recent years, Japan has been actively pursuing the realization of zero carbon cities. However, significant challenges persist, including a lack of effective methods for local governments to communicate zero carbon initiatives to their citizens. This has resulted in limited awareness among citizens about how to participate in zero carbon initiatives. To address these issues, the authors develop a gamified application aimed at promoting zero carbon activities in this research.Through a case study conducted in Sanda City, Hyogo Prefecture in Japan, the authors report the progress of its social implementation.
Zero carbon, zero carbon city, gamification, web application, local government.
Takuya Nakata1, S. Chen2, S. Saiki3, and 2M. Nakamura2, 1Graduate School of Engineering, Kobe University, Rokkodai-cho 1–1, Nada-ku, Kobe, Hyogo, 657–8501 Japan, 2Center of Mathematical and Data Sciences, Kobe University, 1-1 Rokkodai, Nada-ku, Kobe, 657-8501, Japan, 3School of Data and Innovation, Kochi University of Technology, 185 Miyanokuchi, Tosayamada, Kami, Japan
Research concerning the personalization of services encompasses approaches such as machine learning and dialogue agents; however, the explainability of the recommendation process remains a challenge. Previous studies have proposed dialogue-based needs extraction systems utilizing the 6W1H need model, but extracting complex needs using simple natural language processing proved challenging. In this research, we embark on the development of an Application Programming Interface (API) that extracts user needs from natural language by leveraging the rapidly advancing Large Language Models (LLM), and on constructing a dialogue-based needs extraction system using this API. For evaluation, we conducted a verification on 100 needs with the aim of assessing the accuracy and comprehensiveness of the outputs from the needs extraction and restoration API. Through this study, it became feasible to extract needs with high accuracy and comprehensiveness from complex natural language using LLM.
Personalization, need, large language model, natural language processing, dialogue agent.
Fernando W.A.M.A.R, Jinadasa U.G.O.C Amarasinghe B.P.A, Mandalawatta M.T, Uthpala Samarakoon and Manori Gamage, Faculty of Computing, Sri Lanaka Institute of Information Technology, Malabe, Sri Lanaka
The surge in sexually transmitted diseases (STDs) has become a critical public health crisis demanding urgent attention and action. Employing technology to enhance the tracking and management of STDs is vital to prevent their further propagation and to enable early intervention and treatment. This requires adopting a comprehensive approach that involves raising public awareness about the perils of STDs, improving access to affordable healthcare services for early detection and treatment, and utilizing advanced technology and data analysis. The proposed mobile application aims to cater to a broad range of users, including STD patients, recovered individuals, and those unaware of their STD status. By harnessing cutting-edge technologies like image detection, symptom-based identification, prevention methods, doctor and clinic recommendations, and virtual counsellor chat, the application offers a holistic approach to STD management.
STD, Machine Learning, NLP, Neural Network , Image Processing.
H.N Haputhantri, E.G.R.M Piyathissa, W.K.K.H Ariyarathna and G.R.P.S Wijewickrama, Faculty of Computing, Sri Lanka Institute of Information Technology, Malabe, Sri Lanka
The proposed online voting and ranking system for elections in Sri Lanka aims to increase efficiency, reduce time consumption and improve transparency. By leveraging the Internet and digital technology, the proposed system provides a convenient, secure and user-friendly platform for voters to cast their votes from anywhere. Strong authentication methods, such as facial recognition and fingerprint verification, ensure the validity of the voting process and reduce the risk of fraud and manipulation. Voters must verify their identity before voting, and all votes are encrypted to protect voter privacy. The system automatically tabulates and aggregates votes, reducing human error and speeding up the release of results. Real-time updates improve transparency and public confidence in the electoral system. A comprehensive evaluation was conducted to determine the systems effectiveness, including technical framework, security precautions, user approval, and legal considerations. Surveys and interviews were used to assess the usability, safety and fairness of the system. The integration of facial recognition and fingerprint authentication into the proposed online voting and grading system will further enhance its security and reliability. These technologies will help ensure the integrity of the election process and protect the privacy of voters.
Central Counting Station, Voter Registration, Voter Authentication, Data Visualization, Transparent Qualities.
Abderezak Touzene, Ahmed Al Farsi, Nasser Al Zeidi, Department of Computer Science, College of Science, Sultan Qaboos University, Oman
With the emergence of smart devices and the Internet of Things (IoT), millions of users connected to the network produce massive network traffic datasets. These vast datasets of network traffic (Big Data) are challenging to store, deal with and analyse to detect normal or cyber-attack traffic. In this paper we developed an Intrusion Detection System (NMF-IDS) based on Non-Negative Matrix Factorization dimension reduction technique to handle the large traffic datasets and efficiently analyses them in order to detect with a good precision the normal and attack traffic. The experiments we conducted on the proposed IDS-NMF give better results than the traditional ML-based intrusion detection systems, we have got an excellent detection accuracy of 98%.
Intrusion Detection Systems, Machine Learning, Dimensionality Reduction, IoT traffic.
Andrew Chain1, Logan Coons2, Adam Ramos3, Kya Richardson4, Kamaluddeen Usman Danyaro5, Bimal Nepal6, Abdullahi Mujaheed Saleh7, Hashir Sohrab8, 1Department of Computer Science & Engineering, Texas A&M University, Texas, USA 2Department of Chemistry and Biochemistry University of Nevada, Las Vegas, USA, 3Department of Biochemistry & Biophysics, Texas A&M University, Texas, USA 4Department of Bioengineering and Biomedical Engineering, North Carolina A&T State University, North Carolina, USA, 5Department of Computer & Information Science, Universiti Teknologi PETRONAS, Malaysia, 6Department of Engineering Technology & Industrial Distribution, Texas A&M University, Texas, USA, 7Department of Computer & Information Science, Universiti Teknologi PETRONAS, Malaysia, 8Department of Industrial & Systems Engineering, Texas A&M University, Texas, USA
Brain cancer is on the rise globally, with a significant increase in adult brain tumor cases in the last two decades. Detecting and treating brain tumors is challenging due to delayed diagnosis, asymptomatic presentation, and size and shape variations. Gliomas, slow-growing brain tumors, are classified by grade and type. These classifications are useful in predicting the tumors growth rate and likelihood of recurrence. Brain tumors are categorized as benign or malignant. Medical image processing methods can be time-consuming, and accurate grading and typing guidance are scarce. Convolutional neural networks are a deep learning model that can automatically learn and extract notable features from MRI images and is our selected machine learning tool to accomplish accurate classification of brain tumors. It is important to recognize brain tumors early on, so that treatment can be given early in the progression of the disease.
Brain Tumor Classification, Convolutional Neural Network, MRI Images, High-grade Glioma, Image Processing.
Cristian Daniel Alecsa, Technical University of Cluj-Napoca, Cluj-Napoca, Romania, Romanian Institute of Science and Technology, Cluj-Napoca, Romania
In the present paper we introduce new optimization algorithms for the task of density ratio estimation. More precisely, we consider extending the well-known KMM (kernel mean matching) method using the construction of a suitable loss function, in order to encompass more general situations involving the estimation of density ratio with respect to subsets of the training data and test data, respectively. The codes associated to our Python implementation can be found at https://github.com/CDAlecsa/Generalized-KMM.
Kernel mean matching, quadratic optimization, density ratio estimation, loss function.
Mohammed Rashid Al Balushi, Alzahraa Abdullah Alsiyabi, and Yousuf Nasser Al Husaini, Department of Computer Studies, Arab Open University, Muscat, Oman
Drowning is a pressing global issue, responsible for 7% of unintentional injury deaths worldwide. To address this challenge, we proposed An Automated Drowning Rescue Collar, a technologically advanced solution equipped with GPS, accelerometers, and gyroscopes for precise location tracking and depth assessment of individuals in distress. Our research encompasses a comprehensive implementation plan, from design and material acquisition to prototype manufacturing, testing, documentation, and final product development. Powered by solar energy, the Automated Drowning Rescue Collar aligns with Omans sustainability goals and exhibits versatility for various sectors, including aviation, marine, and military applications. The benefits to Oman are substantial, including the potential to save lives, reduce drowning incidents, and support rapid deployment and supply transport. This research harmonizes with Omans vision for technological innovation and sustainable progress. In conclusion, the Automated Drowning Rescue Collar offers a cutting-edge, eco-friendly, and adaptable solution to the global drowning epidemic. Through advanced technology and renewable energy, it holds the promise of safeguarding both rescuers and individuals in peril, thereby making a significant contribution to water safety on a global scale.
Drowning Prevention, Automation, IoT, Water Safety, Technological Innovation, Sustainable Rescue.
Gurleen Kaur, Bakul Gupta, Ashima Anand, Thapar Institute of Engineering and Technology, India
The security of digital images is crucial since they often contain sensitive and confidential data. Unauthorized access to this data could result in severe penalties for the parties involved. Despite the availability of highly secure algorithms, security remains a significant concern due to the rapid emergence of new technologies that can breach it. Thus the proposed work implements a technique that makes the confidential data inaccessible to intruders. Hence fragile type of data hiding technique is used where even with the slightest tampering to the image by an attacker, the information i.e. watermark image is completely destroyed, hence preventing it from unauthorized access. Also, a hybrid transform including DTCWT and NSST is used to fuse two medical images to form a more sophisticated output image, which serves as the final watermark. Further, the zero watermarking model is implemented using the ResNet 50 DL model for more precise results and extraction of feature maps. Embedding the actual image in the carrier image could make the watermarking detectable especially when it is fragile, hence Zero Watermarking overcomes this also by virtual embedding. Moreover, the algorithm employs the avalanche effect of SHA512 for highly secure authentication, further strengthening the security of the system. Overall, the proposed method is an effective way to ensure the security of digital images with confidential data.
Zero watermarking, Image Fusion, RDWT, Encryption, Medical images, Deep Learning.
Josephine (Hsin) Liu1, 2, Phoebe (Yun) Liu1, 2, Joseph (Yu) Liu1, 2, Emily X. Ding1, Robert J. Hou1, 1Vineyards AI Lab,, Auckland, New Zealand, 2Rangitoto College, Auckland, New Zealand
This paper explores the applications of Artificial intelligence (AI) techniques for classifying Deoxyribonucleic Acid (DNA) sequences by us three high school students under teaching and supervision. It presents the processes we did in this research, including investigating DNA sequences, understanding AI models, coding implementation, experiments, and developing a demo for users. During the research, a couple of analogies were introduced to explain AI information and concepts understandably. They were displayed using interesting images to give the high school students a better understanding, and we have successfully achieved our goal of Auto Recognition of DNA Sequences. We first transformed the DNA sequences into human-like language. Then we employed Natural Language Processing (NLP) and Multi-layer perceptron (MLP) to complete sequence classification into 7 gene families from 3 organisms (humans, dogs, and chimpanzees). During this exciting research, we deeply understood the biological and mathematical knowledge we learned in class and adapted them to our research (e.g. DNA-related information for analysis of the experiments, sets, functions, vectors, and matrix, etc., to the classification model). Finally, we used Python and TensorFlow to implement it. The experiments have shown that our project succeeded in achieving high accuracy. In addition, we developed a demo for the user to access the classifier easily.
DNA Sequences, Auto Recognition, Natural Language Processing(NLP), Multi-layer Perceptron (MLP).
Büşra Öztürk1 and Yakup Genç2, 1Computer Engineering Department, Gebze Technical University, Kocaeli, Türkiye, 2Computer Engineering Department, Gebze Technical University, Kocaeli, Turkey
In todays technologies, artificial intelligence systems are frequently used. In these systems, the development phase is as important as the final product. Visualization provides convenience to the user when monitoring data and the model during data collection and development. Augmented reality interfaces offer an effective environment for the user. We often see augmented reality used in various applications alongside artificial intelligence. The use of these interfaces during the development of artificial intelligence systems will provide an immersive experience with a positive impact on the users perception. This study examines research conducted in two dimensions on deep learning models and explores what can be accomplished in three dimensions. By adding augmented reality interfaces as an additional dimension, a three-dimensional experience is provided to the user, and the results are observed. Visualization has been achieved through an application using Microsoft HoloLens 2, and it has been tested with users working with deep learning models to see what its contribution to the user will be in three dimensions.
Deep Learning Models, Visualization, Augmented Reality Interfaces.
Henok Ghebrechristos and Gita Alaghband, Department of Computer Engineering, University of Colorado-Denver, Denver, Colorado
This paper addresses the vulnerability of deep learning models, particularly convolutional neural networks (CNN)s, to adversarial attacks and presents a proactive training technique designed to counter them. We introduce a novel volumization algorithm, which transforms 2D images into 3D volumetric representations. When combined with 3D convolution and deep curriculum learning optimization (CLO), it significantly improves the immunity of models against localized universal attacks by up to 40%. We evaluate our proposed approach using contemporary CNN architectures and the modified Canadian Institute for Advanced Research (CIFAR-10 and CIFAR-100)and ImageNet Large Scale Visual Recognition Challenge (ILSVRC12) datasets, showcasing accuracy improvements over previous techniques. The results indicate that the combination of the volumetric input and curriculum learning holds significant promise for mitigating adversarial attacks without necessitating adversary training.
Convolutional Neural Network, Adversary Attack, Deep Learning,Volumization, Adversary Défense, Curriculum Learning.
Rohit Khankhoje, Independent Researcher Avon, Indiana, US
As the pervasive presence of Smart AI permeates various facets of our technological landscape, the assurance of software quality for AI-driven systems emerges as an imperative of utmost importance. This scholarly paper delves into the intricate ecosystem surrounding Smart AI software, with the objective of unraveling the challenges and issues that are inherent in its quality assurance and articulating the urgent necessity for robust solutions. We undertake an exploration of the multidimensional aspects associated with the challenges in ensuring the quality of Smart AI software. These challenges span from the complexities entailed in handling biased training data to the ethical considerations surrounding the transparency of algorithms. The paper delves into the technical challenges, encompassing the intricacy of testing and the resilience of AI models, and further expands the discussion to encompass societal and ethical considerations, including concerns pertaining to privacy and the establishment of trust among users. Moreover, this paper underscores the compelling need for a comprehensive framework for quality assurance in Smart AI software, with a specific emphasis on its pivotal role in ensuring safety, reliability, and compliance with regulatory standards. The impact of quality assurance on user experience is examined, thereby highlighting the mutually beneficial relationship between the assurance of quality and user satisfaction. By exploring these challenges, issues, and the burgeoning need for effective solutions, this scholarly paper contributes to the ongoing discourse surrounding the responsible development and deployment of Smart AI software, thereby paving the way for advancements in the practices of quality assurance within this dynamic and ever-evolving technological landscape.
Artificial Intelligent, Software Testing, AI Software,Quality Assurance.
Preeti Sharma1, Manoj Kumar2,3,4,5, and Hitesh Kumar Sharma6, 1School of Computer Science, University of Petroleum and Energy Studies (UPES), Dehradun, 248007 India, 2School of Computer Science, FEIS, University of Wollongong in Dubai, Dubai Knowledge Park, Dubai, UAE, 3Research Cluster Head, Network and Cyber Security, UOWD, Dubai 4MEU Research Unit, Middle East University, Amman, 11831, Jordan, 5Research Fellow, INTI International University, Malaysia, 6School of Computer Science, University of Petroleum and Energy Studies (UPES), Dehradun
To produce super-resolution images through GAN, it is essential to eliminate the noise elements and give a clear noise-free output. To achieve this purpose multiscale image representation is found to be effective in many ways for its accuracy of correct feature extraction capacity. This denoising approach is integrated as a chosen enhancement tool in the GAN model, and accordingly, the generator-discriminator training concept is transformed to adopt the approach as per the desired demands. In this research, a multiscale image approach is integrated using an ensemble GAN model with hybrid discriminator architecture. The technique optimises training through simultaneous generator and discriminator model updates, improving output quality, by using the least loss value for discriminator selection. Inception Score (IS) and Fréchet Inception Distance (FID) evaluations show that it outperforms pixel-based denoising, with an amazing accuracy of 99.91%.
GAN, multiscale image representation, ensemble GAN, pixel based denoising, Multiscale denoising.
Amira Ksiksi, Research Groups in Intelligent Machines (REGIM Lab) University of Sfax, National Engineering School of Sfax (ENIS), Sfax, Tunisia
A multi-tenant application aims to provide a single instance of an application with the ca- pability for each organization to have its own specific functionalities. Recent researches have proved the efficiency of the intrusive and non-intrusive approaches in providing deep customizations. However, deep customizations are still limited to the features provided for each organization. In order to enhance the deep customization, we propose a BPMN-based customizations to provide to each organization the capability to create its own features. such a method requires an administration module to provide to the organization to create forms, scripts and notifications to be integrated in a BPMN workflow’s tasks. Such a method has proved its capability to introduce new functionalities using understandable graphical representations which reduce the need for the vendors’ intervention.
Multi-tenancy, BPMN, deep customizations, workflow.
Agbesua Oluwatoyin1, Bello Samuel2, and Lawal Nurudeen3, 1,Department of Insurance with minor in Computer Science, The Polytechnic Ibadan, Nigeria, 2Department of Computer Science, Ladoke Akintola University of Technology, Nigeria, 3Department of Building Technology, Federal Polytechnic of Offa, Nigeria
This article addresses vital challenges in children with Autism Spectrum Disorder (ASD): recognizing emotions and comprehending repetitive behaviors. It employs Software Engineering to create an interactive web app for autism support using autocorrection-based learning. The software focuses on emotion-related queries and caters to ASD childrens preferences. It incorporates animated characters, aligning with research showing their appeal to ASD kids over adult interactions. Interactive elements, like sound effects, so as to create a user-friendly learning environment tailored specifically for children on the Autism Spectrum Disorder. A significant feature is autocorrection, providing constructive feedback through Human-Computer Interaction. Correct answers are highlighted in green, incorrect ones in red, and the correct ones blinked in green, reinforcing emotional understanding. Recognizing ASDs repetition tendency, the software reintroduces previously answered questions, aiding emotional comprehension in line with their learning patterns, as ASD children benefit from repeated information.
Autism Spectrum Disorder (ASD), Software Engineering, Assistive technology, Emotions, Human-Computer Interaction.
Xiuhan Fu1, Theodore Tran2, 1Santa Margarita Catholic High School, 22062 Antonio Pkwy, Rancho Santa Margarita, CA 92688, 2Computer Science Department, California State Polytechnic University, Pomona, CA 91768
This research develops a swimmer’s fitness journal along with a nutritional guidance application. The method employs journaling, expert suggestions, and detailed nutrition tips aimed at tackling precision issues using strictlycontrolled research trials. The application exhibits remarkable accuracy in suggesting proteins and carbohydrates. It promises of more refinement according to personal goals and level of activity. The paper begins by presenting a brief overview of the issue in the swimming community. This highlights the significance of individualized dietary guidelines and coaching services tailored for swimmers. Integration of BMI and BMR calculations for accurate nutritional recommendations is explained in this section of the methodology [13]. The author emphasizes on the need for accuracy in this work. It is also validated to previous standards and methodologies. The section of experiments and results reveals that several experiments were carried out in different situations, which prove the accuracy of the application’s recommendations on nutrition. More nutrients can be incorporated as well. Another part of this article explores possible scenarios where this application applies, including swimmer focused design, more generalized fitness advice and nutrition recommendations targeting a more extensive audience. Existing solutions are compared within a framework of a methodology comparison where one looks at features, effectiveness, and drawbacks. Limitations and avenue of future improvements are outlined in the summary section in addition to extending nutrient recommendations. This has bearing on application’s importance in health of the population. Finally, this detailed fitness diary helps improve swimmer’s training and performance. It might even benefit others who are looking for customized fitness and dietary guidelines.
Swimming, Fitness, Journaling, Advice.
Ledesma. O, Sánchez. M.A, Lamo. P, Escuela Superior de Ingeniería y Tecnología, Universidad Internacional de la Rioja (UNIR), 26004 Logroño, Spain
This paper analyzes the Directed Acyclic Graph (DAG)-based architecture as an alternative to Blockchain technology for critical Internet of Things (IoT) use cases. The speed of transactions and the scalability of Blockchain technology are limitations for critical IoT applications such as vaccine cold chain monitoring. A pilot project has been developed to analyze the speed of the DAG architecture. It simulates monitoring the vaccine cold chain, recording temperatures and alarms. Using the same architecture, two cases with different IoT connectivity technologies in the pilot project are defined: LoRaWAN and Sigfox. The results of these two cases show the comparison between both technologies that show that the DAG architecture can provide the necessary time delays for critical IoT use cases. The main limitation found after the execution of the two cases of the pilot project is related to the need for worldwide coverage of the communications technologies used. For this reason, the study of communications through IoT satellites with global coverage is proposed as future work.
Distributed Ledger Technology, Blockchain, Directed Acyclic Graph, Internet of Things, IOTA.
Tanuj Surve1, Amit Tyagi2, 1University of California, Berkeley, 2National Institute of Fashion Technology
Quadratic Voting has its advantages over other traditional voting systems as Quadratic Voting enables the minority group to show their intensity and preference for a cause. The paper aims to evaluate the efficacy of Quadratic Voting in decentralized governance. The paper uses a qualitative approach and secondary sources to collect the necessary information to serve the purpose of the research. The research highlighted the claimed benefits derived from Quadratic Voting which are not limited to the representation of minority groups, ensuring voter preferences and sincerity in stating preferences, aggregating information efficiently and maximizing social welfare function under assumptions. The paper also sheds light on the limitations and assumptions of Quadratic Voting. Finally, comparative analysis between Quadratic voting and other systems necessitates that the policymakers give their attention to assessing its feasibility and implementation.
Quadrating Voting, Decentralized Governance, Blockchain.
Surya Boppanaa, William Kane, and Long Ma, Department of Computer Science, Troy University, Troy, Alabama, USA
The Secured communication is the ideal approach to communicating with one another without leakage of data. Data encryption is a valuable method for protecting and securing data. This paper proposes novel reversible information, a concealing strategy for computerized pictures using a whole number-to-number wavelet change, and a companding procedure to install and recoup the mystery data and re-establish the image to its perfect state. This paper also presents the use of genetic operators in cryptography. As general information altering happens in the system, the messages should be ensured when transmitted through any system. This paper introduces another encryption method where the genetic operators cross-over and mutation are utilized to encrypt messages to give protection while the information is transmitted. The goal is to provide a better-secured environment by using encryption and a digital image watermarking method for data hiding in the image is paper gives complete guidelines for authors submitting papers for the AIRCC Journals.
Genetic Algorithm, Cryptography, Adaptive Thresholding, Companding Technique, Integer Wavelet Transform, Reversible Watermarking work Protocols.
Lu Yang. and Yan He,School of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing, China
The application of multi-sequence spreading technology in grant-free non-orthogonal multiple access can significantly improve resource utilization and have good adaptability to typical mMTC scenarios in 5G. However, in grant-free multiuser detection through multi-sequence spreading random access, the correlation between spreading sequences directly affects the performance of active user detection. Therefore, this paper proposes a grouping algorithm that can reduce the correlation between different spreading sequence groups. The algorithm performs a low correlation grouping design for the basic spreading sequence set pool based on Gaussian random matrix, further reducing the correlation. Then, compressed sensing technology is used to detect multiple users using low-correlation multi-sequence spreading grouping access. Finally, the algorithm was simulated in a GF-NOMA system based on multiple measurement vector compression sensing, and the results showed that the multiuser detection bit error rate (BER) and error detection probability (DER) of this algorithm were lower than those of other existing multi- sequence grouping methods.
Grant-Free Non-Orthogonal Multiple Access, Multi-Sequence Spreading Random Access, Multiuser Detection, Spreading Sequence Grouping.
Hafiz Bilal Ahmad1 and Fawad Hussain Jaskani2, 1Department of Computer Science and Technology, Xidian University, Xi’an, China, 2Department of Computer Engineering, Islamia University, Bahawalpur, Pakistan
In order to meet the specific requirements of various industries and the stringent demands of 5G, the control and management of 5G networks will heavily depend on the integration of Software Defined Networking, Network Function Virtualization, and Machine Learning. Machine learning can play a crucial role in addressing challenges such as slice type prediction, route optimization, and resource management. To effectively evaluate the use of machine learning in 5G networks, a suitable testing environment is necessary. This study proposes a lightweight testbed that leverages container virtualization technologies to support the development of machine learning net-work functions within 5G networks. The Deep Slice 5G dataset from Kaggle was utilized to predict the type of communication between users based on packet loss and delay budget ratio, with the goal of making 5G systems more efficient. To accomplish this, we applied several Boosted Machine Learning models such as XGBoost, Gradient Boost, AdaBoost, LightGradientBoosting, CatBoost, and HistGradientBoosting. After evaluation, the Catboost model demonstrated the highest accuracy of 99% in identifying the correct slice of 5G based on the selected features of the dataset.
5G Network, Machine Learning, Intrusion Detection System.
Le Ba Nguyen and Ngoc Hong Tran, Computer Science Program, Faculty of Engineering Vietnamese German University, Binh Duong, Vietnam
The Distributed Denial of Service (DDoS) attack is a well-recognized form of cyber attack. A number of approaches and solutions have been devised to detect it. Impressively, data mining methods have been employed to identify patterns of DDoS attacks. Nevertheless, the recent results have not mentioned which factors of the network traffic indicate the potential for attacks. Additionally, with the Machine Learning (ML) approach, there are still opportunities to enhance the detection model accuracy. Furthermore, in this paper, we leverage a variety of ML algorithms for the purpose of improving the accuracy of data classification as ”Benign” or ”DDoS”. The experimental outcomes of our methodologies demonstrate potential and effectiveness.
DDoS, Machine Learning, Random Forest, Naive Bayes, Logistic Regression, MLP.
Akansha Akansha and Stuart So, 1University of Exeter Business School, Rennes Drive, Exeter, United Kingdom
Understanding the user segment is highly significant in the age of a highly competitive wearable Fitness Technology market. In this study, we leveraged a comprehensive dataset containing information on user interactions, activity logs and device usage records. For effective segmentation of the users, K-Means clustering was employed. The unsupervised Machine Learning algorithm helped us group the clusters of consumers based on their similarity in the usage of the device, activity levels and engagement patterns. The collaborative Filtering technique refines product recommendations by identifying user preferences based on past patterns. The analysis aims to uncover distinct user segments and provide insights into user behaviours and lifestyles to enhance Fitbit’s Market Performance and improve user engagement, customer satisfaction and brand loyalty leading to higher customer retention. The findings of an extensive analysis conducted on Fitbit User data using K-Means Clustering and Collaborative filtering techniques are presented. By acknowledging the varying needs of various user segments, Fitbit can enhance its user experience and maintain overall sustainable growth in the extremely competitive market of smart wearables.
Fitbit, Segmentation, K-Means, Collaborative Filtering, Personalisation, Wearable Fitness Technology
Hayastan Avetisyan , Parisa Safikhani, and David Broneske, Department of Research Infrastructure and Methods, DZHW, Hannover, Germany
Authenticating a node in mobile ad-hoc networks is a challenging task due to their dynamic and resource constraint infrastructure. For this purpose, MANETS adopt two kinds of approaches Public key cryptography and identity based cryptography. In Public Key Infrastructure (PKI), Certificate Authority (CA) is responsible for key management. In order to adopt it to MANET, the job of the CA must be distributed. The master secret key is shared among the nodes of the MANET, to self-organize the network without a central authority.The key is shared based on Shamir secret sharing scheme with bi-variate polynomial to make the MANET fully self-managed by nodes.In this paper, we considered PKI based scenario and proposed a new scheme to authenticate a node using BLS signature which is light weight compared to the existing schemes thus making it suitable for MANET.
Mobile ad-hoc network, bi-variate polynomial, secret sharing technique, threshold cryptogra phy, BLS signature
Rohit Khankhoje, Independent Researcher Avon, Indiana, USA
In the constantly evolving world of software development, it is crucial to have ef ective testing methodologies in order to ensure the strength and reliability of applications. This scholarly article presents a new and intelligent approach to test execution that is driven by code and utilizes machine learning to greatly improve adaptability and accuracy in testing processes. Traditional testing methods often struggle to handle changes in code, resulting in less than optimal test execution. Our proposed method utilizes machine learning techniques to predict the impact of code modifications on test results, allowing for a more precise test execution strategy. We have demonstrated significant improvements in test execution ef iciency, reducing unnecessary tests and speeding up feedback cycles. The following discussion examines these findings, addresses potential limitations, and suggests future areas for improvement and expansion. Notably, our methodology explains how Git commits aid in updating features, and how the machine learning model predicts the updated feature names. This predicted feature name is then integrated into Behavior-Driven Development (BDD) test selection and execution using standard BDD frameworks. By seamlessly incorporating machine learning into the testing process, developers can achieve greater precision and ef ectiveness, making significant progress in overcoming challenges posed by changes in code in modern development environments.
Test Automaton, Machine learning, Software testing, Automation Framework,Intelligent Test Strategy.
Licheng Xiao1, Ang Li2, 1Pacific Academy, USA, 2California State Polytechnic University, USA
The problem aimed to solve in this project is a lack of easy to find information and resources in the snowboarding community. Generally, it is common to find information that includes a lack of snowboarding information, resources, how-to-videos, gear item listings and deals, and available resorts when looking for information as a new snowboarder. To solve this problem, we want to make an easy to use mobile app that has informational resources and posts about snowboarding topics as well as listings for great gear items and resorts, including those with current deals or sales. One of the most prominent challenges we faced while developing this app was connecting our FireBase database with the FluttlerFlow application, as our app needed a database to store the gear and resort documents. To connect the FireBase database, we needed to set the corresponding variable in the Firestore within FlutterFlow. In addition, many icons or containers need to have an action that is linked to the firebase and I need to call it in the backend query. To ensure that our FireBase database worked accurately with our FlutterFlow app through the Firestore, we performed tests to ensure that each field in each item in the database was accurate and responded correctly with filters. The result from these tests, we found, was that our filters and fields worked flawlessly, enabling a well-working database set up with our app. Our app is a great solution to the problem we stated because it encompasses a great array of snowboarding related topics and resources, providing an all around informative experience for the user.
Snowboarding, Mobile APP, Firebase, FlutterFlow
Ruijin Deng1, Victor Phan2, 1Shawnigan Lake School, USA, 2California State Polytechnic University, USA
This paper tackles the prevalent problem of students facing resource limitations in pursuing their passions, often due to the lack of like-minded peers within their school environment. To address this issue, we propose the development of a team-management app that serves as a transformative platform, enabling students to connect with others who share their interests and aspirations [1]. The application contains three components: the team management system, the search system, and the chat system. Throughout the development process, we encountered various challenges, such as ensuring user privacy. Comparative analysis with three alternative solutions demonstrated several notable advantages, including an intuitive and user-friendly interface, granting students full autonomy over their clubs, and being open to all students from diverse club interests. This innovative application has the potential to empower students, foster collaboration, and provide a valuable resource for educational institutions and students alike [2].
Students Networking, Natural Language Processing Algorithms, Mobile Application, Social Networking
Yahya BENREMDANE, University Hassan II of Casablanca, Faculté des Sciences Ben M’Sik, B.P. 7955-Sidi Othmane Casablanca, Morocco
This paper aims to find an automatic solution for the modulation’s classification of distinct types of radio signals by relying on Artificial Intelligence. Our work therefore consisted in the choice of the database needed for supervised deep learning, the evaluation of existing techniques on raw communication signals, and the proposal of a solution based on deep learning networks allowing to classify the types of modulation with an optimal ratio (computation time / accuracy). We first conducted a research work on the existing models of automatic classification. We consequently proposed an ensemble learning approach based on tuned ResNet and Transformer Neural Network that is efficient at extracting multi- scale features from the raw I/Q sequence data and considers the challenge of predicting in low Signal Noise Ratio (SNR) conditions. In the end, we delivered an architecture that is easy to manage and apply to communication signals. This solution has an optimal and robust architecture that automatically determines the type of modulation with an accuracy up to 95%.
Automatic modulation classification, Artificial Intelligence, Deep Learning, Radio Frequency.
Copyright © UBIC 2023