Cyberattack detection in wireless sensor networks using a hybrid feature reduction technique with AI and machine learning methods

Journal Of Big Data, Journal Year: 2024, Volume and Issue: 11(1)

Published: Jan. 13, 2024

Abstract This paper proposes an intelligent hybrid model that leverages machine learning and artificial intelligence to enhance the security of Wireless Sensor Networks (WSNs) by identifying preventing cyberattacks. The study employs feature reduction techniques, including Singular Value Decomposition (SVD) Principal Component Analysis (PCA), along with K-means clustering enhanced information gain (KMC-IG) for extraction. Synthetic Minority Excessively Technique is introduced data balancing, followed intrusion detection systems network traffic categorization. research evaluates a deep learning-based feed-forward neural algorithm's accuracy, precision, recall, F-measure across three vital datasets: NSL-KDD, UNSW-NB 15, CICIDS 2017, considering both full reduced sets. Comparative analysis against benchmark approaches also conducted. proposed algorithm demonstrates exceptional performance, achieving high accuracy reliability in WSNs. outlines system configuration parameter settings, contributing advancement WSN security.

Language: Английский

---

Divergent sensory pathways of sneezing and coughing

Cell, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 1, 2024

Language: Английский

---

Weakly-supervised thyroid ultrasound segmentation: Leveraging multi-scale consistency, contextual features, and bounding box supervision for accurate target delineation

Computers in Biology and Medicine, Journal Year: 2025, Volume and Issue: 186, P. 109669 - 109669

Published: Jan. 13, 2025

Language: Английский

---

Artificial intelligence in COVID-19 research: A comprehensive survey of innovations, challenges, and future directions

Computer Science Review, Journal Year: 2025, Volume and Issue: 57, P. 100751 - 100751

Published: April 4, 2025

Language: Английский

---

A multimodal AI-based non-invasive COVID-19 grading framework powered by deep learning, manta ray, and fuzzy inference system from multimedia vital signs

Heliyon, Journal Year: 2023, Volume and Issue: 9(6), P. e16552 - e16552

Published: May 25, 2023

The COVID-19 pandemic has presented unprecedented challenges to healthcare systems worldwide. One of the key in controlling and managing is accurate rapid diagnosis cases. Traditional diagnostic methods such as RT-PCR tests are time-consuming require specialized equipment trained personnel. Computer-aided artificial intelligence (AI) have emerged promising tools for developing cost-effective approaches. Most studies this area focused on diagnosing based a single modality, chest X-rays or cough sounds. However, relying modality may not accurately detect virus, especially its early stages. In research, we propose non-invasive framework consisting four cascaded layers that work together patients. first layer performs basic diagnostics patient temperature, blood oxygen level, breathing profile, providing initial insights into patient's condition. second analyzes coughing while third evaluates imaging data X-ray CT scans. Finally, fourth utilizes fuzzy logic inference system previous three generate reliable diagnosis. To evaluate effectiveness proposed framework, used two datasets: Cough Dataset Radiography Database. experimental results demonstrate effective trustworthy terms accuracy, precision, sensitivity, specificity, F1-score, balanced accuracy. audio-based classification achieved an accuracy 96.55%, CXR-based 98.55%. potential significantly improve speed diagnosis, allowing more control management pandemic. Furthermore, framework's nature makes it attractive option patients, reducing risk infection discomfort associated with traditional methods.

Language: Английский

---

Enhancing Facial Expression Recognition System in Online Learning Context Using Efficient Deep Learning Model

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 121419 - 121433

Published: Jan. 1, 2023

This paper presents an online educational platform that leverages facial expression recognition technology to monitor students' progress within the classroom. Periodically, a camera captures images of students in classroom, processes these images, and extracts data through detection methods. Subsequently, learning statuses are assessed using techniques. The developed approach then dynamically refines enhances teaching strategies acquired status data. In course experiment, we enhance accuracy utilization ResNet-50 for effective feature extraction. Additionally, by adjusting residual down-sampling module, bolster correlation among input features, thus mitigating loss information. Simultaneously, convolutional attention mechanism module is incorporated reduce influence irrelevant areas map. proposed method achieves 87.62% 88.13 % on RAF-DB FER2013 datasets, respectively. comparison with original network outcomes found existing literature, suggested demonstrates enhanced improved states variations. Consequently, application learning, along optimization resources grounded results recognition, holds tangible value augmenting quality experiences. We have benchmarked model against state-of-the-art techniques conducted evaluations FER-2013, CK+, KDEF datasets. significance lies their potential institutions.

Language: Английский

---

Molecular Property Prediction of Modified Gedunin Using Machine Learning

Molecules, Journal Year: 2023, Volume and Issue: 28(3), P. 1125 - 1125

Published: Jan. 23, 2023

Images of molecules are often utilized in education and synthetic exploration to predict molecular characteristics. Deep learning (DL) has also had an influence on drug research, such as the interpretation cellular images well development innovative methods for synthesis organic molecules. Although research these areas been significant, a comprehensive review DL applications would be beyond scope single Account. In this study, we will concentrate major area where influenced design: prediction properties modified gedunin using machine (ML). AI ML technologies critical development. other words, deep algorithms artificial neural networks (ANN) have changed field. short, advances present good potential rational design exploration, which ultimately benefit humanity. paper, long short-term memory (LSTM) was used convert SMILE into image. The 2D representations their immediately visible highlights should then provide adequate data subordinate characteristics atom design. We aim find K-means clustering; RNN-like tools. To support postulation, network (NN) clustering based picture is evaluated study. novel chemical developed via profound predicted As result, LSTM with RNNs allow us gedunin. total accuracy suggested model 98.68%. property promising enough evaluate extrapolation generalization. requires just seconds or minutes calculate, making it faster more effective than existing techniques. can useful tool predicting

Language: Английский

---

Research on Improved DenseNets Pig Cough Sound Recognition Model Based on SENets

Electronics, Journal Year: 2022, Volume and Issue: 11(21), P. 3562 - 3562

Published: Oct. 31, 2022

In order to real-time monitor the health status of pigs in process breeding and achieve purpose early warning swine respiratory diseases, SE-DenseNet-121 recognition model was established recognize pig cough sounds. The 13-dimensional MFCC, ΔMFCC Δ2MFCC were transverse spliced obtain six groups parameters that could reflect static, dynamic mixed characteristics sound signals respectively, DenseNet-121 used compare performance sets optimal set parameters. improved by using SENets attention module enhance model’s ability extract effective features from signals. results showed 26-dimensional MFCC + ΔMFCC, rate accuracy, recall, precision F1 score for sounds 93.8%, 98.6%, 97% 97.8%, respectively. above can be develop a system diseases.

Language: Английский

---

A New Model to Detect COVID-19 Coughing and Breathing Sound Symptoms Classification from CQT and Mel Spectrogram Image Representation using Deep Learning

International Journal of Advanced Computer Science and Applications, Journal Year: 2022, Volume and Issue: 13(8)

Published: Jan. 1, 2022

Deep Learning is a relatively new Artificial Intelligence technique that has shown to be extremely effective in variety of fields. Image categorization and also the identification artefacts images are being employed visual recognition. The goal this study recognize COVID-19 like cough breath noises signals from real-world situations. suggested strategy considers two major steps. first step signal-to-image translation aided by Constant-Q Transform (CQT) Mel-scale spectrogram method. Next, nine deep transfer models (GoogleNet, ResNet18/34/50/100/101, SqueezeNet, MobileNetv2, NasNetmobile) used extract categorise features. digital audio signal will represented recorded voice. CQT transform time-domain input frequency-domain signal. To produce spectrogram, frequency really converted log scale as well colour dimension decibels. construct Mel indeed translated onto scale. dataset contains information over 1,600 people all world (1185 men 415 women). DL model takes spectrograms derived breathing coughing tones patients diagnosed using coswara-combined dataset. With better classification performance employing sound Mel-spectrogram image, current proposal outperformed other CNN networks. For diagnosed, accuracy, sensitivity, specificity were 98.9%, 97.3%, 98.1%, respectively. Resnet18 most reliable network for symptomatic sounds. When applied Coswara dataset, we discovered model's accuracy (98.7%) outperforms state-of-the-art (85.6%, 72.9%, 87.1%, 91.4%) according SGDM optimizer. Finally, research compared comparable investigation. more stable than any present model. Cough precision good enough just test extrapolation generalization abilities. As result, sufferers at their headquarters may utilise novel method main screening tool try identify prioritising patients' RT-PCR testing decreasing chance disease transmission.

Language: Английский

---

Revolutionizing online education: Advanced facial expression recognition for real-time student progress tracking via deep learning model

Multimedia Tools and Applications, Journal Year: 2024, Volume and Issue: unknown

Published: June 5, 2024

Abstract This paper presents a groundbreaking online educational platform that utilizes facial expression recognition technology to track the progress of students within classroom environment. Through periodic image capture and data extraction, employs ResNet50, CBAM, TCNs for enhanced recognition. Achieving accuracies 91.86%, 91.71%, 95.85%, 97.08% on RAF-DB, FER2013, CK + , KDEF datasets, respectively, proposed model surpasses initial ResNet50 in accuracy detection students' learning states. Comparative evaluations against state-of-the-art models using datasets underscore significance results institutions. By enhancing emotion accuracy, improving feature relevance, capturing temporal dynamics, enabling real-time monitoring, ensuring robustness adaptability environments, this approach offers valuable insights educators enhance teaching strategies student outcomes. The combined capabilities contribute uniquely dynamic changes expressions over time, thereby facilitating accurate interpretation emotions engagement levels more effective monitoring behaviors real-time.

Language: Английский

---