Internet of Things (IoT) Security Intelligence: A Comprehensive Overview, Machine Learning Solutions and Research Directions

Published: March 7, 2022

The Internet of Things (IoT) is one the most widely used technologies today, and it has a significant effect on our lives in variety ways, including social, commercial, economic aspects. In terms automation, productivity, comfort for consumers across wide range application areas, from education to smart cities, present future IoT hold great promise improving overall quality human life. However, cyber-attacks threats greatly affect applications environment IoT. traditional security techniques are insufficient with recent challenges considering advanced booming different kinds attacks threats. Utilizing artificial intelligence (AI) expertise, especially machine deep learning solutions, key delivering dynamically enhanced up-to-date system next-generation system. Throughout this article, we comprehensive picture intelligence, which built that extract insights raw data intelligently protect devices against cyber-attacks. Finally, based study, highlight associated research issues directions within scope study. Overall, article aspires serve as reference point guide, particularly technical standpoint, cybersecurity experts researchers working context

Language: Английский

---

Diagnostic Value of Artificial Intelligence-Assisted Endoscopic Ultrasound for Pancreatic Cancer: A Systematic Review and Meta-Analysis

Diagnostics, Journal Year: 2022, Volume and Issue: 12(2), P. 309 - 309

Published: Jan. 25, 2022

We performed a meta-analysis of published data to investigate the diagnostic value artificial intelligence for pancreatic cancer. Systematic research was conducted in following databases: PubMed, Embase, and Web Science identify relevant studies up October 2021. extracted or calculated number true positives, false positives negatives, negatives from selected publications. In total, 10 studies, featuring 1871 patients, met our inclusion criteria. The risk bias included assessed using QUADAS-2 tool. R RevMan 5.4.1 software were used calculations statistical analysis. did not show an overall heterogeneity (I2 = 0%), no significant differences found subgroup pooled sensitivity specificity 0.92 (95% CI, 0.89-0.95) 0.9 0.83-0.94), respectively. area under summary receiver operating characteristics curve 0.95, odds ratio 128.9 71.2-233.8), indicating very good accuracy detection Based on these promising preliminary results further testing larger dataset, intelligence-assisted endoscopic ultrasound could become important tool computer-aided diagnosis

Language: Английский

---

Conditional Generative Adversarial Network Approach for Autism Prediction

Computer Systems Science and Engineering, Journal Year: 2022, Volume and Issue: 44(1), P. 741 - 755

Published: June 1, 2022

Autism Spectrum Disorder (ASD) requires a precise diagnosis in order to be managed and rehabilitated. Non-invasive neuroimaging methods are disease markers that can used help diagnose ASD. The majority of available techniques the literature use functional magnetic resonance imaging (fMRI) detect ASD with small dataset, resulting high accuracy but low generality. Traditional supervised machine learning classification algorithms such as support vector machines function well unstructured semi structured data text, images, videos, their performance robustness restricted by size accompanying training data. Deep on other hand creates an artificial neural network learn make intelligent judgments its own layering algorithms. It takes plentiful low-cost computing many approaches focused very big datasets concerned creating far larger more sophisticated networks. Generative modelling, also known Adversarial Networks (GANs), is unsupervised deep task entails automatically discovering regularities or patterns input for model generate output new examples could have been drawn from original dataset. GANs exciting rapidly changing field delivers promise generative models terms ability realistic across range problem domains, most notably image-to-image translation tasks hasn't explored much spectrum disorder prediction past. In this paper, we present novel conditional adversarial network, cGAN short, which form GAN uses generator conditionally images. accuracy, they outperform standard GAN. proposed 74% accurate than traditional only around 10 min even huge

Language: Английский

---

Leguminous seeds detection based on convolutional neural networks: Comparison of Faster R-CNN and YOLOv4 on a small custom dataset

Artificial Intelligence in Agriculture, Journal Year: 2023, Volume and Issue: 8, P. 30 - 45

Published: April 12, 2023

This paper help with leguminous seeds detection and smart farming. There are hundreds of kinds it can be very difficult to distinguish between them. Botanists those who study plants, however, identify the type seed at a glance. As far as we know, this is first work consider images different backgrounds sizes crowding. Machine learning used automatically classify locate 11 types. We chose Leguminous from types objects study. Those colors, sizes, shapes add variety complexity our research. The dataset was manually collected, annotated, then split randomly into three sub-datasets train, validation, test (predictions), ratio 80%, 10%, 10% respectively. considered variability were captured on five backgrounds: white A4 paper, black pad, dark blue green pad. Different heights shooting angles considered. crowdedness also varied 1 50 per image. combinations arrangements Two image-capturing devices used: SAMSUNG smartphone camera Canon digital camera. A total 828 obtained, including 9801 (labels). contained backgrounds, heights, angles, crowdedness, arrangements, combinations. TensorFlow framework construct Faster Region-based Convolutional Neural Network (R-CNN) model CSPDarknet53 backbone for YOLOv4 based DenseNet designed connect layers in convolutional neural. Using transfer method, optimized models. currently dominant object methods, R-CNN, performances compared experimentally. mAP (mean average precision) R-CNN models 84.56% 98.52% had significant advantage speed over which makes suitable real-time identification well where high accuracy low false positives needed. results showed that better accuracy, ability, faster beating by large margin. effectively applied under image levels It constitutes an effective efficient method detecting complex scenarios. provides reference further testing enumeration applications.

Language: Английский

---

Using Social Media Analysis to Improve E-commerce Marketing Strategies

International Review of Management and Marketing, Journal Year: 2024, Volume and Issue: 14(4), P. 61 - 71

Published: July 5, 2024

This study investigates the application of game theory and matrix-based analysis in enhancing social media marketing strategies for e-commerce businesses. By integrating these mathematical models with analytics, aim to provide a comprehensive framework that can predict consumer behavior, optimize competitive strategies, improve engagement on digital platforms. study's matrix model showcased clear benefits entities, aggressive tactics boosting market share by 30% against passive competitors achieving 20% increase even when also adopted approaches. The Nash Equilibrium emphasize balanced gains both firms engaged strategies. Statistical reinforced efficacy chi-square test yielding significant value 13.4, suggesting strong link between enhanced metrics. Regression further validated impact sales, indicating 1% likes, comments, shares corresponded 0.75% uplift evidenced predictors β values 0.25, 0.35, 0.40 respectively. Content surveys highlighted preference authentic, value-aligned content, leading 50% higher rate 60% such emphasizing critical role strategic alignment expectations. Incorporating into offers novel approach understanding leveraging complex interplay interactions dynamics. methodology enables marketers devise more targeted, adaptive, effective campaigns, driving growth satisfaction marketplace.

Language: Английский

---

Enhancing Urban Sustainability: Developing an Open-Source AI Framework for Smart Cities

Smart Cities, Journal Year: 2024, Volume and Issue: 7(5), P. 2670 - 2701

Published: Sept. 18, 2024

To address the growing need for advanced tools that enable urban policymakers to conduct comprehensive cost-benefit analyses of traffic management changes, Urbanite H2020 project has developed innovative artificial intelligence methods. Among them is a robust decision support system assists in evaluating and selecting optimal mobility planning modifications by combining objective subjective criteria. Utilising open-source microscopic simulation tools, accurate digital models (or “digital twins”) four pilot cities—Bilbao, Amsterdam, Helsinki, Messina—were created, each addressing unique challenges. These challenges include reducing private vehicle access Bilbao’s city center, analysing impact increased bicycle population growth constructing mobility-enhancing tunnel improving public transport connectivity Messina. The research introduces five key innovations: application consistent platform across diverse environments, integration consistency challenges; pioneering use Dexi smart cities; implementation visualisations; machine learning tool, Orange, with user-friendly GUI interface. innovations collectively make complex data analysis accessible non-technical users. By applying multi-label techniques, decision-making process accelerated three orders magnitude, significantly enhancing efficiency. project’s findings offer valuable insights into both anticipated unexpected outcomes interventions, presenting scalable, AI-based framework decision-makers worldwide.

Language: Английский

---

Investigating the effect of fuel properties and environmental parameters on low-octane gasoline-like fuel spray combustion and emissions using machine learning-global sensitivity analysis method

Energy, Journal Year: 2024, Volume and Issue: 306, P. 132551 - 132551

Published: July 22, 2024

Language: Английский

---

A Comprehensive Review to Understand the Definitions, Advantages, Disadvantages and Applications of Machine Learning Algorithms

International Journal of Computer Applications, Journal Year: 2024, Volume and Issue: 186(31), P. 43 - 47

Published: July 30, 2024

Machine learning (ML) means that first the machine learns with help of algorithms then works automatically.In today's age people want to do almost everything automatically and efficiently.In sense has made a revolutionary change because its efficiency.An intelligent faster than human.The incidence errors is conspicuously decreased by using ML.Depending on improving necessity ML in present situation this paper tried describe some especially supervised, unsupervised, semi-supervised reinforcement including their definitions, advantages, disadvantages area work so will understand which algorithm where use.Particularly Support Vector (SVM), Decision Trees, K-Nearest Neighbors (K-NN), Linear Regression, Logistic Regression for supervised learning.K-Means Clustering, Principal Component Analysis (PCA) unsupervised learning.Basics learning.Eventually from can easily get idea commonly used algorithms.

Language: Английский

---

Intelligent credit scoring using deep learning methods

Concurrency and Computation Practice and Experience, Journal Year: 2023, Volume and Issue: 35(9)

Published: Feb. 13, 2023

Summary Credit scoring is one the most important parts of credit risk management in reducing client defaults and bankruptcies. Deep learning has received much attention recent years, but it not been implemented so intensively compared to other financial domains. In this article, stacked unidirectional bidirectional LSTM (long short‐term memory) networks as a complex area deep are applied solving problems for first time. The proposed robust model exploits full potential three‐layer BDLSTM (bidirectional LSTM) architecture with treatment modeling public datasets novel way since time sequence problem. Attributes each loan instance were transformed into matrix fixed sliding window approach one‐time step. Our models outperform existing more solutions thus we succeeded preserving simplicity. measures different types employed carry out consistent conclusions. results by applying three hidden layers on German dataset showed an accuracy 87.19%, Kaggle reached 93.69%, Microcredit 97.80%.

Language: Английский

---

Industrial applications of software defect prediction using machine learning: A business-driven systematic literature review

Information and Software Technology, Journal Year: 2023, Volume and Issue: 159, P. 107192 - 107192

Published: March 8, 2023

Machine learning software defect prediction is a promising field of engineering, attracting great deal attention from the research community; however, its industry application tents to lag behind academic achievements. This study part larger project focused on improving quality and minimising cost testing 5G system at Nokia, aims evaluate business applicability machine gather lessons learnt. The systematic literature review was conducted journal conference papers published between 2015 2022 in popular online databases (ACM, IEEE, Springer, Scopus, Science Direct, Google Scholar). A quasi-gold standard procedure used validate search, SEGRESS guidelines were for transparency, reporting, replicability. We have selected analysed 32 publications out 397 found by our automatic search (and seven snowballing). identified highly relevant evidence methods, features, frameworks, datasets used. However, we minimal emphasis practical learnt consciousness — both vital perspective. Even though number studies validated increasing able identify several excellent performed vivo), there still not enough focus aspects effort that would help bridge gap needs research.

Language: Английский

---