Scholarly Indexed Publications

Permanent URI for this collectionhttps://hdl.handle.net/10361/28605

Browse

Now showing 1 - 20 of 1491

Metadata Only
Detecting faulty machinery of waste water treatment plant using statistical analysis & machine learning
(Institute of Electrical and Electronics Engineers Inc., 2022-01-01) Ul Islam, Md. Mazed; Mondal, Joyanta Jyoti; Shihab I.F.; Department of Computer Science and Engineering
The goal of wastewater treatment is to eliminate contaminants from wastewater and convert them into effluent/discharge that can be reintroduced into the water cycle. In order to monitor, analyze plant performance, and decrease environmental pollution in wastewater treatment facilities, a model for fault detection must be developed. In this study, we examine different time and cost-efficient machine learning approaches to monitor the operation of a Waste Water Treatment Plant (WWTP) and identify plant faults as an alternative to human, laboratory-based time consuming, costly, and challenging techniques. This will allow us to develop a time and cost-efficient approach to detect such problems. To discover plant defects, we collect one year of unsupervised WWTP data and convert the data into supervised data. Using several machine learning algorithms based on water quality standard measurements (pH, BOD, COD, and suspended solid), we establish whether or not the data is valid.
Metadata Only
LSTM-ANN based price hike sentiment analysis from Bangla social media comments
(Institute of Electrical and Electronics Engineers Inc., 2022-01-01) Chakraborty, Sovon; Uddin Talukdar, Muhammad Borahn; Yaseen Morshed Adib, Muhammed; Mitra, Sowmen; Rabiul Alam, Md. Golam; Department of Computer Science and Engineering
Price hike has always been a substantial concern for people all over the world. The crisis gets more conspicuous, and people find themselves more confounded when even the bare minimum of expenses still exceeds the amount they can get to earn. This tension tends to invite chaos in society as the number of people affected increases. Bangladesh is currently undergoing a formidable wave of price hikes. People have been expressing mixed reactions on social media regarding this issue. Hence, understanding the overall public sentiment can be crucial for policymaking and preventing chaos in society. This study utilizes social media comments for analyzing underlying sentiments. Data were collected from the Facebook pages of some popular Bangladeshi media for this purpose, and thereby a specialized dataset was constructed. The dataset contains 2000 public comments annotated with three polarity values- positive, negative, and neutral. A hybrid LSTM-ANN deep architecture has been exploited in this research. The model outperforms other state-of- the-art models in terms of less trainable parameters along with an F1-score of 88.47%.
Metadata Only
A data-driven insight to enhancing stress management through chatbot interaction among undergraduate students
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Zaman, Md. Mohibur; Rahman, Md. Rifat; Ur Rasul, Mehbub Shifat; Zaman, Md Showrav; Fahim, Md. Mohyminul Islam; Miami, Anika Tahsin; Sakif, Md. Sadiqul Islam; Noor, Jannatun; Department of Computer Science and Engineering
Student stress management at the undergraduate level is a significant issue in the educational world. Therefore, this stress is a challenging issue that needs to be dealt with. As higher educational requirements stack up and their challenges grow, most students struggle to keep the required balance between their studies and preparation. This study establishes a deeper investigation of stress management among undergraduate students using machine-learning algorithms to identify factors contributing to stress and provide solutions. The research aims to illuminate the fundamental causes and health implications of stress for students. Through surveys and questionnaires, the study categorizes stress stages, identifying patterns and enabling the use of new stress management strategies. The findings are used to address concerns shared by undergraduate students and determine interventions to help them cope effectively which aim to provide students with the strength, knowledge, resources, and support needed to manage stress effectively and live a fulfilling life during their university years and future years. The findings will be used to develop policies and programs backed by science to ensure emotional and academic success for students.
Metadata Only
Interpretable deep learning approaches for reliable GI image classification: A study with the HyperKvasir dataset
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Wahid, Saif Bin; Rothy, Zarin Tasnim; News, Raisul Kabir; Rieyan S.A.; Department of Computer Science and Engineering
Deep learning has emerged as a promising tool for automating gastrointestinal (GI) disease diagnosis. However, multi-class GI disease classification remains underexplored. This study addresses this gap by presenting a framework that uses advanced models like InceptionNetV3 and ResNet50, combined with boosting algorithms (XGB, LGBM), to classify lower GI abnormalities. InceptionNetV3 with XGB achieved the best recall of 0.81 and an F1 score of 0.90. To assist clinicians in understanding model decisions, the Grad-CAM technique, a form of explainable AI, was employed to highlight the critical regions influencing predictions, fostering trust in these systems. This approach significantly improves both the accuracy and reliability of GI disease diagnosis.
Metadata Only
FAANG stock price prediction: A hybrid approach integrating deep learning with ensemble learning
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Anis, Sadaf M.; Kabbya, Md Asif Shahidullah; Talukder, Anika Hasan; Chowdhury, Iffat Jahan; Hossain, Muhammad Iqbal; Department of Computer Science and Engineering
In this research, a novel hybrid approach has been proposed by implementing several well known deep learning models such as : Long Short-Term Memory (LSTM), Artificial Neural Network (ANN), Autoencoder, Multilayer Perceptron (MLP), and Recurrent Neural Network (RNN) and integrating the deep learning models with ensemble learning techniques such as - Stacking Ensemble, Voting ensemble to a unified framework to predict stock prices of major FAANG companies dataset. Proposed ensemble learning techniques give better performance and accuracy than individual deep learning models. The stacking ensemble uses a meta-learner to combine predictions from individual deep learning models, while on the other hand voting ensemble averages the predictions. Stacking ensemble demonstrates consistent performance across datasets from major FAANG companies outperforming individual models and reducing the variability observed in standalone predictions. The research also employs Advanced Evaluation Metrics such as : R2 score and Mean Squared Logarithmic Error (MSLE) , F1 - score to evaluate performance providing nuanced insights into predictive accuracy for time-series data with exponential trends. Unlike traditional methods that rely only on a single model our proposed hybrid approach leverages complementary strengths of multiple architectures ensuring robustness and improved accuracy across datasets. By integrating predictions through a meta learner the proposed method achieves consistent performance, outperforming individual models. The novelty of this research lies in the comprehensive integration of multiple deep learning models and regularization techniques which ensure enhanced generalization and overcoming the inherent challenges in time-series financial forecasting across various datasets.
Metadata Only
Generalizing cancer lesion detection from limited data using YOLOv8 and transfer learning
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Muntasir, Fahim; Akter T.; Datta A.; Quaium M.A.; Department of Computer Science and Engineering
Automatically detecting different skin cancers from an image of a skin lesion can greatly help medical professionals in early diagnosis. It can also aid in non-invasive skin cancer identification. However, the lack of dataset availability, bias, class imbalance and suitable ways to work with a small sample of data are poorly defined areas for skin cancer identification. This research addresses these issues with a novel algorithm incorporating tuning image augmentation and training YOLOv8 in a transfer learning manner. A dataset with 1000 images of five different skin cancers was collected and annotated for cancer detection with the YOLOv8 algorithm. The base YOLOv8 was first trained for a larger epoch to create a baseline model, and the trained model was stripped to retrain with optimized hyperparameters. The final model performed closely to other models trained on large datasets. Evaluating unseen images with our trained model confirmed its applicability to real-life scenarios. This study demonstrates the effectiveness of using this method in improving the YOLOv8's detection performance and providing a more effective solution to multi-class skin cancer detection from small sample sizes.
Metadata Only
Automated identification of ocular toxoplasmosis in fundoscopic images utilizing deep learning models
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Zaman, Arsi; Choudhury, Prionto Kumar; Chowdhury, Rizvee Rifat; Anika, Asma Akter; Rahman Ramisa, Sumaiya; Department of Computer Science and Engineering
The detection of ocular toxoplasmosis is based on eye fundus images, which are produced and analysed by a specialist optician. Although Deep Learning techniques are being quickly adopted in many areas, their application in ocular diagnosis has not been widely studied. In this research, we created a highly effective Convolutional Neural Network (CNN) model that can accurately identify and classify ocular Toxoplasmosis (OT) images into four categories: healthy, active, inactive, and active-inactive. After further examination, we merged three of these categories - active, inactive, and active-inactive - into a single class called "unhealthy,"resulting in a binary classification of healthy and unhealthy images. Our ensemble model shows 96% accuracy. Our CNN model shows a high level of accuracy in distinguishing between these two categories. To validate the performance of our custom CNN model, we compared it with three pre-trained models (VGG16, VGG19, and MobileNet) using the same dataset. The findings revealed that both our proposed CNN model and the pre-trained architectures displayed comparable performance metrics, including accuracy, recall, F1 score, and precision. Our model achieved a remarkable accuracy of 97%, surpassing the performance of previously utilised models in diagnosing retinal disorders.
Open Access
TritonZ: A remotely operated underwater rover with manipulator arm for exploration and rescue operations
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Ahmed K.; Fardin M.S.; Nayem M.A.F.; Hafiz F.; Shatabda, Swakkhar; Department of Computer Science and Engineering
The increasing demand for underwater exploration and rescue operations enforces the development of advanced wireless or semi-wireless underwater vessels equipped with manipulator arms. This paper presents the implementation of a semi-wireless underwater vehicle, "TritonZ"equipped with a manipulator arm, tailored for effective underwater exploration and rescue operations. The vehicle's compact design enables deployment in different submarine surroundings, addressing the need for wireless systems capable of navigating challenging underwater terrains. The manipulator arm can interact with the environment, allowing the robot to perform sophisticated tasks during exploration and rescue missions in emergency situations. TritonZ is equipped with various sensors such as Pi-Camera, Humidity, and Temperature sensors to send real-time environmental data. Our underwater vehicle controlled using a customized remote controller can navigate efficiently in the water where Pi-Camera enables live streaming of the surroundings. Motion control and video capture are performed simultaneously using this camera. The manipulator arm is designed to perform various tasks, similar to grasping, manipulating, and collecting underwater objects. Experimental results shows the efficacy of the proposed remotely operated vehicle in performing a variety of underwater exploration and rescue tasks. Additionally, the results show that TritonZ can maintain an average of 13.5cm/s with a minimal delay of 2-3 seconds. Furthermore, the vehicle can sustain waves underwater by maintaining its position as well as average velocity. The full project details and source code can be accessed at this link: https://github.com/kawser-ahmed-byte/TritonZ
Metadata Only
DeepTransit: Optimizing urban package delivery with conveyor network simulation
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Helal M.R.; Chowdhury M.A.; Gilman K.; Chowdhury, Sabiha Alam; Ghimire S.; Chowdhury M.A.; Department of Computer Science and Engineering
This paper presents the development and analysis of simulation software designed to model an underground package delivery network utilizing conveyor belts in cities such as New York. The proposed system integrates with New York's subway network, addressing urban logistical challenges by efficiently handling large volumes of package, reducing costs, time, and environmental impact. The software simulates the operation of this system and considering factors such as delivery time, distance traveled, package delays, and heuristic efficiency. By executing experimental simulation techniques, mathematical models were constructed to relate various simulation parameters, resulting in an effective heuristic cost function for package routing.
Metadata Only
Enhancing object detection interpretability for class-discriminative visualizations with Grad-CAM
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Ishrak, Md Fatin; Nahar, Jannatun; Faiza, Fairooz Afnad; Siddiqua, Lamia Mahzabin; Department of Computer Science and Engineering
Computer vision has benefited greatly from deep learning; it has saved the day by making it possible to achieve marvelous performances in visual recognition challenges. In the field of object detection, not only the detection of objects in video and image data but also the localization of those objects is essential. This work proposes a new way to perform object detection based on Grad-CAM with the main emphasis on the ILSVRC-17 dataset. A higher level of generality made it possible to adapt Grad-CAM originally designed for representing the image classification to highlight action regions in the frames of the video and images. Our visualizations offer several advantages: ig they are useful in the following ways: (a) They offer guidance on model failure modes and deepen our understanding of how seemingly irrational predictions are logical (b) They perform effectively and correctly on the ILSVRC-17 weakly supervised localization task compared to previous methodologies (c) They demonstrate resilience against adversarial disturbances (d) They are closer to the true model in their representations and (e) They help the models to generalize. The proposed work shows that the idea of Grad-CAM helps in increasing action localization effectiveness and, at the same time, provides meaningful information about the model's performance and decision-making mechanisms. These results demonstrate that Grad-CAM can be an efficacious tool for action recognition and spatial prediction, which lays the foundation for video and image processing, surveillance, and human-machine interaction. Lastly, we also talk about the possibilities of refining object detection approaches and increasing the model's interpretability, which we believe is important for the field of computer vision.
Metadata Only
Deepfake detection using neural networks
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Asha, Wajida Anwar; Saba, Nures; Huq, Syed Mahbubul; Hossain, Muhammad Iqbal; Department of Computer Science and Engineering
The widespread use of deepfake technology, which uses advanced artificial intelligence techniques like Generative Adversarial Networks (GANs), poses severe threats to public confidence and digital security. This study investigates sophisticated methods for identifying deepfake content, emphasising the Convolutional Neural Networks (CNNs) ensemble, namely the ResNeXt and EfficientNetB4 models. Using the Celeb-DF (v2) dataset, which includes more than 5,600 deepfake videos, we provide a unique method that combines Siamese training techniques and attentional mechanisms inside CNN structures. With our improvements, EfficientNetB4 could identify edited films with an Area Under Curve (AUC) score of 99% and an exceptional accuracy of 97%. These findings highlight the potential of improved CNN models to differentiate between complex deepfakes and authentic videos accurately. Future research directions include expanding dataset diversity and applying transfer learning to refine detection techniques further, thereby contributing to the secure dissemination of digital content.
Metadata Only
Q-learning based automated message multicast in gossip protocol for node confirmation in IOTA tangle
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Keya, Mahmuda; Mandal, Shovon; Dhar, Swarojani ; Tahsin, H.M. ; Tahsin, Anika; Siam, Morshed; Alam, Md. Golam Rabiul; Uddin M.Z.; Department of Computer Science and Engineering
The Internet of Things (IoT) is now causing a massive wave of digitization, creating vast amounts of data in the Internet of Everything era. Distributed ledger technologies like blockchains and IOTA are essential to IoT data services. The IOTA Foundation has completely redesigned distributed ledger technology to enable secure cash and data exchange, fee-free microtransactions, and scalable network growth for IoT device networks. It provides an approach and transaction confirmations to enable smart device microtransactions. The quicker the network uses the IOTA Tangle, the more transactions are verified. However, IOTA's maximum transaction rate of approximately seven transactions per second (TPS) limits its potential. In May 2020, the community network achieved 600 confirmed Transactions Per Second (CTPS), showcasing progress. Motivated by the need to enhance IoT scalability and efficiency, this research proposes a methodology to integrate Tangle into IoT blockchains, leveraging Tangle as the backbone for IoT devices. The approach addresses message flooding, reducing overhead while offering a distinct web interface cost-cutting strategy to minimize transaction time and storage for microtransactions. Experimental results demonstrate that the framework improves transaction throughput, substantially reducing processing time and resource usage. These findings underscore its efficacy in enabling efficient and scalable IoT microtransactions. The research concludes that the proposed integration of Tangle enhances IoT transaction efficiency and sets a foundation for future innovations in secure, lightweight IoT data exchange.
Open Access
Deep learning for breast cancer detection: Comparative analysis of ConvNeXT and EfficientNet
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Hasan, Mahmudul; Department of Computer Science and Engineering
Organizational success in today's competitive employment market depends on choosing the right staff. This work evaluates software engineer profiles using an automated staff selection method based on advanced natural language processing (NLP) techniques. A fresh dataset was generated by collecting LinkedIn profiles with important attributes like education, experience, skills, and self-introduction. Expert feedback helped transformer models - including RoBERTa, DistilBERT, and a customized BERT variation, LastBERT - to be adjusted. The models were meant to forecast if a candidate's profile fit the selection criteria, therefore allowing automated ranking and assessment. With 85% accuracy and an F1 score of 0.85, RoBERTa performed the best; DistilBERT provided comparable results at less computing expense. Though light, LastBERT proved to be less effective, with 75% accuracy. The reusable models provide a scalable answer for further categorization challenges. This work presents a fresh dataset and technique as well as shows how transformer models could improve recruiting procedures. Expanding the dataset, enhancing model interpretability, and implementing the system in actual environments will be part of future activities.
Metadata Only
Automated personnel selection for software engineers using LLM-based profile evaluation
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Jawad Karim, Ahmed Akib; Hoque, Shahria; Alam, Md. Golam Rabiul; Uddin M.Z.; Department of Computer Science and Engineering
Organizational success in today's competitive employment market depends on choosing the right staff. This work evaluates software engineer profiles using an automated staff selection method based on advanced natural language processing (NLP) techniques. A fresh dataset was generated by collecting LinkedIn profiles with important attributes like education, experience, skills, and self-introduction. Expert feedback helped transformer models - including RoBERTa, DistilBERT, and a customized BERT variation, LastBERT - to be adjusted. The models were meant to forecast if a candidate's profile fit the selection criteria, therefore allowing automated ranking and assessment. With 85% accuracy and an F1 score of 0.85, RoBERTa performed the best; DistilBERT provided comparable results at less computing expense. Though light, LastBERT proved to be less effective, with 75% accuracy. The reusable models provide a scalable answer for further categorization challenges. This work presents a fresh dataset and technique as well as shows how transformer models could improve recruiting procedures. Expanding the dataset, enhancing model interpretability, and implementing the system in actual environments will be part of future activities.
Metadata Only
Optimizing endpoint detection and monitoring in enterprise solution: A cyber threat intelligence approach
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Sarker, Apurba; Mondal, Joty Prokash; Seraj, Mehnaz; Noor, Jannatun; Department of Computer Science and Engineering
Advanced cyber threat intelligence systems are crucial at a time when enterprise solutions are increasing and are being targeted by more sophisticated cyberattacks. This research aims to improve endpoint detection and monitoring in enterprises by installing a Security Information and Event Management (SIEM) system based on Wazuh having an open-source robust and flexible Host Intrusion Detection System (HIDS) with integration into ELK Stack. The infrastructure creates a powerful combination of analytical security, intrusion detection, log data analysis, file integrity monitoring (FIM), and vulnerability management capabilities using the Wazuh active response module to detect security threats and continuous monitoring. Wazuh provides a robust and cost-effective solution that instantly detects and monitors simulated attacks such as denial-of-service (DoS) attacks by spotting suspicious file changes in real-time, SSH authentication failure, and identifying the root source of the flood of requests. This study also provides useful insights on designing and deploying comprehensive cybersecurity solutions with open-source tools such as Wazuh, making visual insights for file integrity monitoring (FIM) in real time.
Metadata Only
Efficient action detection in video sequences: A hybrid CRNN-transformer approach with attention and HPC strategies
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Ishrak, Md Fatin; Hossain, Mohammad Ahad; Mitra, Anindya; Araf, Sheikh Abdul; Rahman, Mushfiqur; Faroque, Md Omar; Department of Computer Science and Engineering
Due to the digital processing advancement and availability of big data the field of action recognition had a great growth in recent years. This work introduces a new action detection model that combines a hybrid CRNN-Transformer model with current computational methods. The proposed model architecture presents a shift from mainstream models, although it provides superior levels of accuracy and performance. One of these is the use of the attention mechanism for focusing selectively, the second stage focuses on object identification and proposing the regions of interest, and the third uses optimized High-Performance Computing methods in the process. The model adopts the highly effective Transformer-based architecture, applied to the spatiotemporal sequences of videos in the form of 5D tensors using ResNet-164 as a base network, thus enhancing the recognition capacities. Moreover, there are Long Short-Term Memory (LSTM) layer that helps capture dependency across time, so predictive chance will increase as well. The proposed hybrid model is better compared to the current approaches; the results of performance analysis on the UCF101 dataset yielded an mAP of 50.1% and a mean class-wise AP (mcAP) of 80.8% which outperforms TSN, LRCN, I3D, and all the other previous models. Other additional techniques are Mixed Precision Training, XLA Optimization, and Gradient Checkpointing to enhance the computational speed. Evaluation measures, including Average Precision (AP), confirm the efficiency of the proposed method in all aspects of action detection.
Metadata Only
Know your chemistry set: Exploring chemical laboratory object detection using YOLOv8 and YOLOv9
(Institute of Electrical and Electronics Engineers Inc., 2024-01-01) Siddiqui J.H.; Ahmed R.U.; Ashrafi A.F.; Arafin, Sumiya; Department of Computer Science and Engineering
Image object detection (IOD) has proven its usefulness from diagnosing critical diseases from medical image analysis to pedestrian recognition in autonomous vehicle tracking. Considering the potential applications of object detection in real-life scenarios, various deep learning-based algorithms have been used in recent years. However, one unexplored sector of object detection is its application in critical environments like a chemical lab. Automatic apparatus/chemical reagent/machine detection can lessen the effect of chemical hazards in these environments as well as can be used to ensure efficient usage of laboratory resources. In this study, the potential of you only look once (YOLO) has been explored for the detection of chemical apparatus from a comprehensive image dataset. The study was validated against a dataset of 5078 images containing 7 different most commonly used chemistry laboratory apparatus that are used chemical laboratory. Our experimentation demonstrates state-of-the-art performance on the detection of objects with an impressive mAP of 0.818 and 0.865 using YOLOv8 and YOLOv9 architectures respectively.
Metadata Only
A machine learning approach to predict movie success from youtube trailer comments
(Institute of Electrical and Electronics Engineers Inc., 2022-01-01) Khan, Farden Ehsan; Ruhan, Ahmed Mahir; Shamsuddin, Rifat; Ashraf, Faisal Bin; Department of Computer Science and Engineering
Social media use has increased to such levels in recent years that it has transformed into a trend-setting powerhouse, introducing subjects that would have previously remained outside of the public eye. Through people's shared opinions and responses about a trend on social media, we hope to determine how long it can hold an audience's attention on its own. We will analyze the sentiment of individuals toward a particular topic using the information gleaned from social media comments. Our work will be based on unreleased films and make predictions about how they will turn out when they are released. In this work, we have processed and examined accumulated reviews about a film to see whether the general public feels positively or negatively about it and to calculate the likelihood that a certain film will be a success. From this, we can infer how the success of a movie or product is influenced by both positive and negative attention before its release.
Open Access
Ethics and methods for collecting sensitive data: Examining sexual and reproductive health needs of and services for Rohingya refugees at Cox’s bazar, Bangladesh
(University of Hawaii at Manoa, 2020-01-01) Aktar, Bachera; Ahmed, Rushdia; Hassan, Raafat; Farnaz, Nadia; Ray, Pushpita; Awal, Abdul; Shafique, Sharid Bin; Hasan, Md Tanvir; Quayyum, Zahidul; Jafarovna M.B.; Kobeissi L.H.; El Tahir K.; Chawla B.S.; Rashid S.F.; BRAC James P Grant School of Public Health
During humanitarian emergencies, such as the forced displacement of the Rohingya diaspora, women and adolescent girls become highly vulnerable to sexual and reproductive health (SRH) issues and abuse. Although sensitive in nature, community-driven information is essential for designing and delivering effective community-centric SRH services. This article provides an overview of the theoretical framework and methodologies used to investigate SRH needs, barriers, and challenges in service-delivery and utilization in the Rohingya refugee camps in Cox’s Bazar, Bangladesh. It also offers insights on important methodological and ethical factors to consider while conducting research in a similar context. A concurrent mixed-method study was undertaken in ten randomly selected Rohingya refugee camps between July and November 2018. The design consisted of a cross-sectional household survey of 403 Rohingya adolescent girls and women, along with an assessment of 29 healthcare facilities. The team also completed in-depth interviews with nine adolescent girls, 10 women, nine formal and nine informal healthcare providers, key informant interviews with seven key stakeholders and seven influential community members. Lastly, three focus group discussions were undertaken with a group of 18 Rohingya men. Our theoretical framework drew from the socio-ecological models developed by Karl Blanchet and colleagues (2017) insofar as they considered a multiplicity of related contextual and cross-cutting factors. Building good rapport with community gatekeepers was key in accessing and sustaining the relationship with the various respondents. The data collected through such context-specific research approaches is critical in designing community-centric service-delivery mechanisms, and culturally and gender-sensitive SRH interventions in humanitarian crises.
Metadata Only
IoT network attack detection using XAI and reliability analysis
(Institute of Electrical and Electronics Engineers Inc., 2022-01-01) Tabassum, Sabrina; Parvin, Nazia; Hossain, Nigah; Tasnim, Anika; Rahman, Rafeed; Hossain, Muhammad Iqbal; Department of Computer Science and Engineering
IoT has emerged as one of the most sophisticated techniques in recent years. But inadequate security controls are the most typical barrier to IoT expansion as the devices transmit a huge amount of data. Nowadays different machine learning and deep learning models are used to detect various IoT attacks. In our previous research, we applied Decision Tree, Random Forest, AdaBoost, XGBoost, ANN, and MLP to the IoT/IIoT dataset of TON IoT datasets to classify IoT network attacks. In binary classification, we got above 96% accuracy for all methods. In contrast, AdaBoost and ANN underperformed in multiclass classification. As accuracy improves, models get more complicated, and these models are frequently seen as black boxes that are difficult to interpret. Though these models give highly precise results, an explanation is required in order to comprehend and accept the models' decisions. Here comes XAI which emphasizes a variety of ways for breaking the black-box nature of Machine Learning and Deep Learning models as well as delivering human-level explanations. In this article, we have extended our work by analyzing different machine learning and deep learning methodologies using XAI to explain the categorization of IoT network attacks. LIME, SHAP, and ELI5 approaches have been used to interpret and explain which will increase transparency and reliability.

Browse

Recent Submissions