Articles

Access the latest knowledge in applied science, electrical engineering, computer science and information technology, education, and health.

Advanced Search
Filter Icon

Filters article

Years

FAQ Arrow
Single Year
Range Year
0
0

Source Title

FAQ Arrow

Authors

FAQ Arrow

31,456 Article Results

Ensemble model-based arrhythmia classification with local interpretable model-agnostic explanations

10.11591/ijai.v14.i3.pp2012-2025
Md. Rabiul Islam , Tapan Kumar Godder , Ahsan Ul-Ambia , Ferdib Al-Islam , Anindya Nag , Bulbul Ahamed , Nujhut Tanzim , Md. Estiak Ahmed
Arrhythmia can lead to heart failure, stroke, and sudden cardiac arrest. Prompt diagnosis of arrhythmia is crucial for appropriate treatment. This analysis utilized four databases. We utilized seven machine learning (ML) algorithms in our work. These algorithms include logistic regression (LR), decision tree (DT), extreme gradient boosting (XGB), K-nearest neighbors (KNN), naïve Bayes (NB), multilayer perceptron (MLP), AdaBoost, and a bagging ensemble of these approaches. In addition, we conducted an analysis on a stacking ensemble consisting of XGB and bagging XGB. This study examines various arrhythmia detection techniques using both a single base dataset and a composite dataset. The objective is to identify the optimal model for the combined dataset. This study aims to evaluate the efficacy of these models in accurately categorizing normal (N) and abnormal (A) heartbeats as binary classes. The empirical findings demonstrated that the stacking ensemble approach exhibited superior accuracy when used with the combined dataset. Arrhythmia classification models rely on this as a crucial component. The binary classification achieved an accuracy of 98.61%, a recall of 97.66%, and a precision of 97.77%. Subsequently, the local interpretable model-agnostic explanations (LIME) technique is employed to assess the prediction capability of the model.
Volume: 14
Issue: 3
Page: 2012-2025
Publish at: 2025-06-01

Designing high power efficient finite impulse response filters with three-four inexact adder-integrated Booth multiplier

10.11591/ijra.v14i2.pp204-213
Manju Inasu Kollannur , Oudaya Coumar Souprayen
Finite impulse response (FIR) filters are widely utilized in several applications in digital signal processing, including data transmission, photography, digital audio, and biomedicine. It is necessary to use high sample rates for FIR filters, while moderate sample rates are needed for low-power circuits. To solve these problems, a Booth multiplier based on three-four inexact adder-based multiplication (TFIE-BM) was proposed. The goal of the proposed TFIE-based FIR Booth multiplier is to lower area usage, latency, and power consumption. The proposed method utilizes the spotted hyena optimizer (SHO) to find the optimal filter coefficient (FC) by minimizing the pass power consumption and Transition bandwidth. Moreover, a high-performance three-four inexact adder (TIFE adder) has been introduced, which uses fewer XOR gates for sum and carry generation, indicating that the logic has been simplified to reduce hardware complexity. By increasing speed and decreasing the FIR filter's critical path delay, a modified Booth multiplier that uses a 5:2 compressor is introduced. The overall delay of the proposed approach is 23.4%, 18.7%, 12.3%, and 5.7% lower than that of the Radix-4 Booth multiplier, CSA Booth multiplier, hybrid multiplier, and traditional Booth multiplier, respectively.
Volume: 14
Issue: 2
Page: 204-213
Publish at: 2025-06-01

Rooftops detection with YOLOv8 from aerial imagery and a brief review on rooftop photovoltaic potential assessment

10.11591/ijai.v14.i3.pp2282-2290
Md. Sabbir Ahmed , Md. Shohel Arman , Nusrat Tasnim , Md Hafizul Imran , Musabbir Smmak , Touhid Bhuiyan
Recent years have seen significant advancements in the switch from fossil fuel-based energy systems to renewable energy. Decentralized solar photovoltaic (PV) is one of the most promising energy sources since there is a lot of rooftop space, it is easy to install, and the cost of the PV panels is low. The determination of rooftop locations for PV installation is crucial for energy planning. With this context, this study aimed to detect the suitable rooftops of different shapes. The dataset of 5,076 building roofs used in this study was gathered by us utilizing a drone. This study identified ten distinct roof shapes accurately, including triangle, square, penta, hexa, hepta, octa, nona, deca, gabled roof, and hipped roof, using the most recent version of you only live once (YOLO), known as YOLOv8. Recent research revealed, YOLOv8 is more accurate than earlier YOLO models which is the reason of utilizing YOLOv8. Accuracy of this work of rooftops detection is 93.6%. Also, the precision, recall, and F1-score confidence curve showed good performances too. Finally, a brief review of the most recent studies on the evaluation of rooftop PV potential was conducted to provide insight into the use of solar energy.
Volume: 14
Issue: 3
Page: 2282-2290
Publish at: 2025-06-01

NAPLAM: a novel ledger-based algorithm for detection and mitigation of sinkhole attacks in routing protocol for low power and lossy networks-based Internet of things

10.11591/ijra.v14i2.pp248-259
Akshaya Dhingra , Vikas Sindhu , Lakshay Dhingra
The Internet of Things (IoT) is a network of connected physical objects that collect and share data over the Internet. However, routing attacks can disrupt data exchange, especially multi-node sinkhole attacks in low power and lossy IoT networks (LLNs). To support communication in LLN IoT, the IPv6-based routing protocol for LLNs (RPL) is used. Despite having several advantages, RPL also faces challenges like being vulnerable to attacks, having limited resources, compatibility, and scalability issues. Additionally, traditional security methods often do not work well for LLN-IoT devices because they lack the necessary computing power. To overcome these challenges, we have proposed a novel ledger-based framework called network and packet ledger to ascertain malicious devices using routing protocol for LLN (NAPLAM-RPL). This framework can effectively detect and mitigate multi-node sinkhole attacks in IoT networks. This paper also compares NAPLAM-RPL with similar protocols using the NetSim Simulator. The experimental analysis shows that NAPLAM-RPL improves network performance and outperforms existing methods like RF-trust, SoS-RPL, INTI, C-TRUST, and heartbeat algorithm in crucial areas, including packet delivery rate (PDR), throughput, End-to-End (E2E) delay, energy consumed, and detection accuracy.
Volume: 14
Issue: 2
Page: 248-259
Publish at: 2025-06-01

Optimizing real-time data preprocessing in IoT-based fog computing using machine learning algorithms

10.11591/ijai.v14.i3.pp1900-1909
Nandini Gowda Puttaswamy , Anitha Narasimha Murthy
In the era of the internet of things (IoT), managing the massive influx of data with minimal latency is crucial, particularly within fog computing environments that process data close to its origin. Traditional methods have been inadequate, struggling with the high variability and volume of IoT data, which often leads to processing inefficiencies and poor resource allocation. To address these challenges, this paper introduces a novel machine learning-driven approach named real-time data preprocessing in IoT-based fog computing using machine learning algorithms (IoT-FCML). This method dynamically adapts to the changing characteristics of data and system demands. The implementation of IoT-FCML has led to significant performance enhancements: it reduces latency by approximately 0.26%, increases throughput by up to 0.3%, improves resource efficiency by 0.20%, and decreases data privacy overhead by 0.64%. These improvements are achieved through the integration of smart algorithms that prioritize data privacy and efficient resource use, allowing the IoT-FCML method to surpass traditional preprocessing techniques. Collectively, the enhancements in processing speed, adaptability, and data security represent a substantial advancement in developing more responsive and efficient IoT-based fog computing infrastructures, marking a pivotal progression in the field.
Volume: 14
Issue: 3
Page: 1900-1909
Publish at: 2025-06-01

Effectiveness of artificial intelligence-driven chatbot responses in diabetes knowledge: a readability and reliability assessment

10.11591/ijai.v14.i3.pp2379-2388
Muhammad Thesa Ghozali , Woro Supadmi , Fadya Bella Suci Maharani , Aloina Jean Rassyifa
Patient education is vital in diabetes management, empowering patients with necessary knowledge and skills to manage condition effectively. However, traditional educational methods often face challenges such as limited access to healthcare professionals and variability in information quality. This study aimed to assess the reliability and readability of artificial intelligence (AI)-driven chatbot responses in disseminating diabetes knowledge. Technically, the diabetes knowledge questionnaire (DKQ-24) was administered to evaluate the effectiveness of AI-driven chatbot in disseminating diabetes-related information. Responses were evaluated for reliability and quality applying the modified DISCERN (mDISCERN) scale and global quality scale (GQS), and readability was assessed using the Flesch reading ease (FRE) score, Flesch-Kincaid grade level (FKGL), gunning fog index (GFI), Coleman-Liau index (CLI), and simple measure of gobbledygook (SMOG). The mean mDISCERN score was 31.50±2.89, indicating generally reliable responses. The median GQS score was 4, reflecting the high overall quality. The readability assessment revealed a mean FRE score of 66.30, indicating that the text was fairly easy to read. FKGL mean score was 6.54±3.19, suggesting the text was suitable for readers at a sixth-grade level. In conclusion, AI-driven chatbot provides reliable and high-quality information on the diabetes self-management, but it requires improvements to enhance accessibility.
Volume: 14
Issue: 3
Page: 2379-2388
Publish at: 2025-06-01

The influence of sentiment analysis in enhancing early warning system model for credit risk mitigation

10.11591/ijai.v14.i3.pp1829-1838
Angel Karentia , Derwin Suhartono
One important source of bank income is interest income from credit activities, another part of which is obtained from fee-based income. Rapid credit growth is directly proportional to an increase in potential credit risk (counterparty default). In addition to comprehensive credit assessment at the initial stage of credit initiation, banks need to monitor the condition of existing debtors. Empirically, difficulties in handling non-performing loans often occur due to delays in detection and preparation of action plans. In this case, losses due to non-performing loans can have implications for the bank's reputation and worsen its financial performance. This research aims to determine the effect of sentiment analysis (external sentiment prediction model [positive, neutral, and negative] with certain keywords) on the level of accuracy of the early warning system (EWS) model in predicting the credit quality of bank debtors in the coming months. This study found that upgrading EWS with sentiment analysis will give better accuracy levels compared to traditional EWS models. In addition, the predictive power of EWS (traditional and upgraded) is inversely proportional to the prediction period, the longer the target prediction time, and the less predictive power of the EWS model.
Volume: 14
Issue: 3
Page: 1829-1838
Publish at: 2025-06-01

Experimental evaluation of bidirectional encoder representations from transformers models for de-identification of clinical document images

10.11591/ijra.v14i2.pp273-280
Ravichandra Sriram , Siva Sathya Sundaram , S. LourduMarie Sophie
Many health institutes maintain patients’ diagnosis and treatment reports as scanned images. For healthcare analytics and research, large volumes of digitally stored patient information have to be accessed, but the privacy requirements of protected health information (PHI) limit the research opportunities. Particularly in this artificial intelligence (AI) era, deep learning models require large datasets for training purposes, which hospitals cannot share unless the PHI fields are de-identified. Manual de-identification is beyond possible, with millions of patient records generated in hospitals every day. Hence, this work aims to automate the de-identification of clinical document images utilizing AI models, particularly pre-trained bidirectional encoder representations from transformers (BERT) models. For the purpose of experimentation, a synthetic dataset of 550 clinical document images was generated, encompassing data obtained from diverse patients across multiple hospitals. This work presents a two-stage transfer learning approach, initially employing Tesseract character recognition (OCR) to convert clinical document images into text. Subsequently, it extracts PHI fields from the text for de-identification. For the purpose of extraction, BERT models were utilized; in this work, we contrasted six pre-trained versions of such models to examine their effectiveness and achieve the F1 score of 92.45%, thus showing better potential for de-identifying PHI data in clinical documents.
Volume: 14
Issue: 2
Page: 273-280
Publish at: 2025-06-01

Skin cancer diagnosis using hybrid deep pre-trained convolutional neural networks

10.11591/ijai.v14.i3.pp2291-2301
Maha Ibrahim Khaleel , Amir Lakizadeh
As a variant of skin cancer, melanoma represents a substantial menace to the health and overall well-being of individuals. Statistics reveal that 55% of skin cancer patients succumb to this particular disease. However, early detection plays a crucial role in reducing mortality rates and saving lives. In the past several decades, there has been a rise in the adoption of deep learning algorithms, capturing the interest of researchers working in this field. One popular method involves utilizing pre-trained deep neural networks. In this study, a hybrid approach is employed to extract features from melanoma images. This approach integrates the utilization of pre-trained architectures, including AlexNet, ResNet-50, and GoogleNet. During the transfer training phase, these networks are fine-tuned to detect skin cancer by adjusting the learning rate. Subsequently, the maximum relevance minimum redundancy (MRMR) algorithm is employed to select optimal features based on the concepts of minimum redundancy and maximum relevance in order to minimize feature redundancy and enhance classification accuracy. The bagging technique is employed for the classification of various skin cancer types. The experimental results demonstrate the success of the suggested approach, yielding 98.9% accuracy. Furthermore, the results indicate the superiority of this method according to precision, recall, and F1-score in comparison with existing algorithms.
Volume: 14
Issue: 3
Page: 2291-2301
Publish at: 2025-06-01

Addressing the challenges of harmonizing law and artificial intelligence technology in modern society

10.11591/ijai.v14.i3.pp2471-2478
Lamprini Seremeti , Sofia Anastasiadou , Andreas Masouras , Stylianos Papalexandris
The invasion of artificial intelligence (AI) in all forms of human activity causes the sudden change of social cohesion into a new hybrid reality, where the static rule of law maybe is overthrown by the instant violations of fundamental human rights, including the general rights of personhood, in its image, honor, and privacy, as well as, of general principles of law, including the principle of the “abuse of rights”, the principle of contractual autonomy, principles of tort liability, and general principles of intellectual property law. In that sense, AI disrupts the acquis due to the poor regulatory quality indicators covering unforeseen occurrences. We call this instantiation AI legal “coup d’état”. This paper constitutes a philosophical thesis statement which is in accordance with the global efforts to legally embed AI into societal systems. As part of an ongoing research on AI and law synergy, this paper focuses on proposing a theoretical framework utilizing category theory to align AI functionalities with traditional legal principles.
Volume: 14
Issue: 3
Page: 2471-2478
Publish at: 2025-06-01

A new era of technological change in the restaurant industry: focusing on perceived values of robot servers

10.11591/ijra.v14i2.pp143-150
Jinsoo Hwang , Kyuhyeon Joo , Joonho Moon
The objective of this research is to examine the perceived values of robot servers, which include utilitarian and hedonic values, and how this influences willingness to pay more in the restaurant industry. This paper also examined the differences between the two sub-dimensions of perceived value, which are based on the demographic factors of the respondents. This research performed a data analysis based on a sample size of 295 participants, and the results indicated that the two sub-dimensions of perceived value play a crucial role in regard to the formation of willingness to pay more. Furthermore, the results showed that there were differences in perceived value in regard to the demographic factors.
Volume: 14
Issue: 2
Page: 143-150
Publish at: 2025-06-01

Multilayer stacking for polycystic ovary syndrome diagnosis

10.11591/ijai.v14.i3.pp1968-1975
Kazi Abu Taher , Samia Ahmed , Jannatul Ferdous Esha , Md. Sazzadur Rahman , A. S. M. Sanwar Hosen
Polycystic ovary syndrome (PCOS) is a complicated hormonal condition that is experienced by women. Despite extensive research, the precise reason be hind PCOS remains unknown, and effective treatments are still lacking. Thus, early diagnosis and treatment have a significant positive impact on the health of women. Recently, there has been remarkable performance demonstrated by machine learning (ML)-based detection models for PCOS identification. They are fast and low cost compared to the traditional processes. In this work, a multi stacking PCOS detection model is proposed using K-fold cross validation. The model uses three different ML algorithms namely: na¨ıve Bayes (NB), ran dom forest (RF), and logistic regression (LR) as base classifiers and a neural network, multi-layer perception (MLP) as meta model. This approach utilizes two feature selection techniques and compares the performances on the stack ing methods. Among the two feature selection techniques, Pearson correlation approach performed better with average 98.79% accuracy, 99.17% sensitivity, 98.40% specificity, and 98.79% f1-score.
Volume: 14
Issue: 3
Page: 1968-1975
Publish at: 2025-06-01

Classification of morphologically similar Indian rice variety using machine learning algorithms

10.11591/ijece.v15i3.pp3202-3211
Harini Shadaksharappa , Saritha Chakrasali , Krishnamurthy Gorappa Ningappa
India, among the agriculture-based economy grows wide variety of rice along with other crops. These varieties have different commercial values as they are different in their features. It becomes extremely challenging to classify rice varieties which have similar features but are different in their quality. This study considers four varieties of similar looking rice which conform to be Sona-Masuri. A total of 4180 images are considered to extract 56 features including textural, red, green, blue (RGB) and hue, saturation, value (HSV) color and wavelet decomposition. Machine learning (ML) models like support vector machine (SVM), K-nearest neighbor (KNN), decision tree (DT) and voting classifiers are developed for feature dataset and convolutional neural network (CNN) model for image dataset. The results obtained for every model are obtained using statistical methods and the results are expressed in a table for accuracy, precision, recall and F1-score. A classification accuracy of 72.48% is obtained for SVM using polynomial kernel trick by considering all 56 features. The customized CNN model is designed with three convolution layers has resulted in 97.13% of training accuracy and 87.5% of validation accuracy. Based on the results obtained, it is witnessed that the ML models employed in this study to classify rice types with similar appearances have practical applications.
Volume: 15
Issue: 3
Page: 3202-3211
Publish at: 2025-06-01

Semantic based medical visual question answering with explainable artificial intelligence

10.11591/ijai.v14.i3.pp2169-2177
Sheerin Sitara Noor Mohamed , Kavitha Srinivasan , Raghuraman Gopalsamy
The medical visual question answering (MVQA) system takes the advantage of both computer vision (CV) and natural language processing (NLP) to accept the medical image and corresponding question as input and generates the respective answer as output. One step further, the MVQA system capable of generating the answer based on the semantics has a distinct place and hence semantic based medical visual question answering (SMVQA) system is proposed in this research. In SMVQA, the semantics for input image and question are generated using layerwise relevance propagation explainable artificial intelligence (LRP XAI) technique and the answer is derived using deductive reasoning method. For this, seven MVQA datasets are used for model creation, testing and validation. The training phase of the SMVQA system is implemented using VGGNet, long short-term memory (LSTM), LRP XAI, ResNet and bidirectional encoder representations from transformers (BERT) to generate a model file. Then the inference is derived in the testing phase based on the generated model file for the test set. Finally, the answer is derived from the inference using natural language toolkit (NLTK) library, term frequency-inverse document frequency (TF-IDF), cosine similarity, best match25 (BM25) techniques along with deductive reasoning. As a result, the proposed SMVQA system gives improved performance then the existing MVQA system especially for abnormality type samples.
Volume: 14
Issue: 3
Page: 2169-2177
Publish at: 2025-06-01

Ledger on internet of things: a blockchain framework for resource-constrained devices

10.11591/ijai.v14.i3.pp2506-2518
Suresh Jaganathan , Karthika Veeramani
The increasing use of resource-constrained devices such as the internet of things (IoT) in various applications has led to the need for an optimized blockchain framework for these devices. Blockchain-based IoT networks allow businesses to access and share IoT data within their organization without centralized authority. However, existing frameworks are not designed for IoT applications and lack features like decentralization, scalability, and network overhead. To overcome these limitations, a new blockchain framework is proposed: ledger on internet of things (LIoT), which has a new consensus-based leader election algorithm to address the challenges of existing algorithms with high block creation time and communication overhead. Moreover, a novel data structure has been developed to reduce the storage size of the ledger effectively. The proposed framework also employs a docker for deployment, which provides an efficient and easy setup of blockchain nodes without requiring the individual configuration of each machine, increases the efficiency of the consensus process, and enables convenient deployment and management of the blockchain framework on resource-constrained devices. Furthermore, the performance of the proposed consensus method is analyzed using various performance parameters, including CPU usage, memory usage, transaction execution time, and block generation time.
Volume: 14
Issue: 3
Page: 2506-2518
Publish at: 2025-06-01
1 2 3
...
2097 2098
Show 293 of 2098

Discover Our Library

Embark on a journey through our expansive collection of articles and let curiosity lead your path to innovation.

Explore Now
Library 3D Ilustration