International Journal of Intelligent Systems and Applications @ijisa
Journal articles - International Journal of Intelligent Systems and Applications
All articles: 1264
Data Mining of Students’ Performance: Turkish Students as a Case Study
Scientific article
Artificial neural networks have been used in different fields of artificial intelligence, and more specifically in machine learning. Although, other machine learning options are feasible in most situations, but the ease with which neural networks lend themselves to different problems which include pattern recognition, image compression, classification, computer vision, regression etc. has earned it a remarkable place in the machine learning field. This research exploits neural networks as a data mining tool in predicting the number of times a student repeats a course, considering some attributes relating to the course itself, the teacher, and the particular student. Neural networks were used in this work to map the relationship between some attributes related to students’ course assessment and the number of times a student will possibly repeat a course before he passes. It is the hope that the possibility to predict students’ performance from such complex relationships can help facilitate the fine-tuning of academic systems and policies implemented in learning environments. To validate the power of neural networks in data mining, Turkish students’ performance database has been used; feedforward and radial basis function networks were trained for this task. The performances obtained from these networks were evaluated in consideration of achieved recognition rates and training time.
Free
Data Quality for AI Tool: Exploratory Data Analysis on IBM API
Scientific article
A huge amount of data is produced in every domain these days. Thus for applying automation on any dataset, the appropriately trained data plays an important role in achieving efficient and accurate results. According to data researchers, data scientists spare 80% of their time in preparing and organizing the data. To overcome this tedious task, IBM Research has developed a Data Quality for AI tool, which has varieties of metrics that can be applied to different datasets (in .csv format) to identify the quality of data. In this paper, we will be representing how the IBM API toolkit will be useful for different variants of datasets and showcase the results for each metrics in graphical form. This paper might be found useful for the readers to understand the working flow of the IBM data purifier tool, thus we have represented the entire flow of how to use IBM data quality for the AI toolkit in the form of architecture.
Free
Scientific article
About one person dies every minute from cardiovascular disease; consequently, it has almost surpassed war as the largest cause of death in the twenty-first century. In cardiology, early and accurate diagnosis of heart illness is a cornerstone of effective healthcare. Predictive analytics, which involves machine-learning algorithms, can be a great option for contributing towards the early detection of cardiovascular disease. This study evaluates the data preprocessing techniques involved in building machine learning models to predict cardiovascular disease and identify the features contributing to the cardio attack. A novel data transformation technique named the superlative boundary binning method was proposed to enhance machine learning and ensemble learning classification models for predicting cardiac illness based on independent physiological feature parameters. The results revealed that the ensemble learning classifier AdaBoost using the superlative boundary binning method has performed well with a classification accuracy of 93% when compared with the other data transformation and machine learning classifier models.
Free
Data Visualization and its Proof by Compactness Criterion of Objects of Classes
Scientific article
In this paper considered the problem of reducing the dimension of the feature space using nonlinear mapping the object description on numerical axis. To reduce the dimensionality of space used by rules agglomerative hierarchical grouping of different - type (nominal and quantitative) features. Groups do not intersect with each other and their number is unknown in advance. The elements of each group are mapped on the numerical axis to form a latent feature. The set of latent features would be sorted by the informativeness in the process of hierarchical grouping. A visual representation of objects obtained by this set or subset is used as a tool for extracting hidden regularities in the databases. The criterion for evaluating the compactness of the class objects is based on analyzing the structure of their connectivity. For the analysis used an algorithm partitioning into disjoint classes the representatives of the group on defining subsets of boundary objects. The execution of algorithm provides uniqueness of the number of groups and their member objects in it. The uniqueness property is used to calculate the compactness measure of the training samples. The value of compactness is measured with dimensionless quantities in the interval of [0, 1]. There is a need to apply of dimensionless quantities for estimating the structure of feature space. Such a need exists at comparing the different metrics, normalization methods and data transformation, selection and removing the noise objects.
Free
Scientific article
This paper proposes a data-driven approximation of the Cumulative Distribution Function using the Finite Mixtures of the Cumulative Distribution Function of Logistic distribution. Since it is not possible to solve the logistic mixture model using the Maximum likelihood method, the mixture model is modeled to approximate the empirical cumulative distribution function using the computational intelligence algorithms. The Probability Density Function is obtained by differentiating the estimate of the Cumulative Distribution Function. The proposed technique estimates the Cumulative Distribution Function of different benchmark distributions. Also, the performance of the proposed technique is compared with the state-of-the-art kernel density estimator and the Gaussian Mixture Model. Experimental results on κ−μ distribution show that the proposed technique performs equally well in estimating the probability density function. In contrast, the proposed technique outperforms in estimating the cumulative distribution function. Also, it is evident from the experimental results that the proposed technique outperforms the state-of-the-art Gaussian Mixture model and kernel density estimation techniques with less training data.
Free
Scientific article
In this research, we established a machine learning–based model to predict the suitability of tsunami evacuation locations in Padang City through the Extreme Gradient Boosting (XGBoost) method. We trained the model on a new synthetic dataset with 5,000 observations with key geospatial and demographic features such as elevation, distance to coastline, suggested evacuation capacity, surrounding population count and site area. The analysis process consisted of preprocessing, feature selection utilizing the XGBoost Classifier, training and cross-validation on each model, and evaluation through regression as well as classification metrics. The XGBoost model performed best (RMSE=0.0642, MAE=0.0418 and Accuracy=93.8%), which was even better than Random Forest, Gradient Boosting Trees and Logistic Regression models. These findings demonstrate that XGBoost can successfully extract complicated spatial–demographic associations with little overfitting. The residual analysis and the actual-vs-predicted plots also reveal good model calibration and stability. A web prototype was also created to visualize the suitability of evacuation and facilitate spatial decision making. Although the model is based on simulated data, it offers an extendible and interpretable framework to be integrated in practical scenarios with field and operational disaster management systems. To the best of our knowledge, this work represents the first use of XGBoost algorithm in Indonesia to classify tsunami evacuation sites and functions as a new tool for disaster preparedness and evacuation plans on the coast.
Free
Scientific article
Accurate electricity forecasting is vital for grid stability and effective energy management, particularly in regions like Benghazi, Libya, which face frequent load shedding, generation deficits, and aging infrastructure. This study introduces a data-driven framework to forecast electricity load, generation, and deficits for 2025 using historical data from two distinct years: 2019 (an instability year) and 2023 (a stability year). Various time series models were employed, including Autoregressive Integrated Moving Average (ARIMA), seasonal ARIMA, dynamic regression ARIMA, extreme gradient boosting, simple exponential smoothing, and Long Short-Term Memory (LSTM) neural networks. Data preprocessing steps—such as missing value imputation, outlier smoothing, and logarithmic transformation—are applied to enhance data quality. Model performance was evaluated using metrics such as mean squared error, root mean squared error, mean absolute error, and mean absolute percentage error. LSTM outperformed other models, achieving the lowest mentioned metric values for forecasting load, generation, and deficits, demonstrating its ability to handle non-stationarity, seasonality, and extreme events. The study’s key contribution is the development of an optimized LSTM framework tailored to North Benghazi’s electricity patterns, incorporating a rich dataset and exogenous factors like temperature and humidity. These findings offer actionable insights for energy policymakers and grid operators, enabling proactive resource allocation, demand-side management, and enhanced grid resilience. The research highlights the potential of advanced machine learning techniques to address energy-forecasting challenges in resource-constrained regions, paving the way for a more reliable and sustainable electricity system.
Free
Scientific article
It is quite common to have access to geospatial (temporal/spatial) panel data generated by a set of similar data for analyses in a meta-data setup. Within this context, researchers often employ pooling methods to evaluate the efficacy of meta-data analysis. One of the simplest techniques used to combine individual-study results is the fixed-effects model, which assumes that a true-effect is equal for all studies. An alternative, and intuitively-more-appealing method, is the random-effects model. A paper was presented by the first author, and his co-authors addressing the efficient estimation problem, using this method in the aforesaid meta-data setup of the ‘Geospatial Data’ at hand, in Map World Forum meeting in 2007 at Hyderabad; INDIA. The purpose of this paper had been to address the estimation problem of the fixed-effects model and to present a simulation study of an efficient confidence-interval estimation of a mean true-effect using the panel-data and a random-effects model, too in order to establish appropriate ‘confidence interval’ estimation for being readily usable in a decision-makers’ setup. The present paper continues the same perspective, and proposes a much more efficient estimation strategy furthering the gainful use of the ‘Geospatial Panel-Data’ in the Global/Continental/ Regional/National contexts of “Socioeconomic & other Developmental Issues’. The ‘Statistical Efficient Confidence Interval Estimation Theme’ of the paper(s) has a wider ambit than its applicability in the context of ‘Socioeconomic Development’ only. This ‘Statistical Theme’ is, as such, equally gainfully applicable to any area of application in the present world-order at large inasmuch as the “Data-Mapping” in any context, for example, the issues in the topically significant area of “Global Environmental Pollution-Mitigation for Arresting the Critical phenomenon of Global Warming”. Such similar issues are tackle-able more readily, as the impactful advances in the “GIS & GPS” technologies have led to the concept of “Managing Global Village” in terms of ‘Geospatial Meta-Data’. This last fact has been seminal to special zeal-n-motivation to the authors to have worked for this improved paper containing rather a much more efficient strategy of confidence-interval estimation for decision-making team of managers for any impugned area of application.
Free
Scientific article
Social media’s worldwide expansion over the past two decades has significantly altered the dissemination of extremist narratives, creating both challenges and opportunities for counterterrorism efforts. Addressing critical gaps in the detection and classification of extremist content on social media platforms, this research supports earlier-stage analytical assessment for law enforcement and security agencies. Using datasets from the publicly available Global Terrorism Database (GTD, n > 209,000 incidents) and a curated corpus of labeled tweets (n = 17,410), a hybrid framework integrating machine learning and deep learning models through a late-fusion stacking architecture is developed. The proposed ensemble leverages contextual indicators derived from historical terrorism data alongside linguistic and behavioral signals from social media content to distinguish extremist from non-extremist activity. Evaluated under strict temporal validation, the model achieves an accuracy of 98.52%, precision of 97.01%, recall of 99.66%, and an AUC of 0.92 under controlled experimental conditions. To address ethical and transparency considerations, Shapley Additive exPlanations (SHAP) are employed to enhance collectively indicate that integrating interpretability in automated decision-making. While the reported results reflect dataset-specific evaluation, the findings historical terrorism intelligence with temporally ordered social media analysis can support counterterrorism efforts by mitigating digital radicalization pathways and associated downstream physical security risks linked to terrorism and extremism through earlier analytical intervention.
Free
Scientific article
In the paper the deep hybrid system of computational intelligence with architecture adaptation for medical fuzzy diagnostics is proposed. This system allows to increase a quality of medical information processing under the condition of overlapping classes due to special adaptive architecture and training algorithms. The deep hybrid system under consideration can tune its architecture in situation when number of features and diagnoses can be variable. The special algorithms for its training are developed and optimized for situation of different system architectures without retraining of synaptic weights that have been tuned at previous steps. The proposed system was used for processing of three medical data sets (dermatology dataset, Pima Indians diabetes dataset and Parkinson disease dataset) under the condition of fixed number of features and diagnoses and in situation of its increasing. A number of conducted experiments have shown high quality of medical diagnostic process and confirmed the efficiency of the deep hybrid system of computational intelligence with architecture adaptation for medical fuzzy diagnostics.
Free
Deep Learning Based Traffic Management in Knowledge Defined Network
Scientific article
In recent Artificial Intelligence developments, large datasets as knowledge are a prime requirement for analysis and prediction. To manage the knowledge of the network, the Data Center Network (DCN) has been considered a global data storage facility on edge servers and cloud servers. In recent research trends, knowledge-defined networking (KDN) architecture is considered, where the management plane works as the knowledge plane. The major network management task in the DCN is to control traffic congestion. To improve network management, i.e., optimized resource management, enhanced Quality of Service (QoS), we propose a path prediction technique by combining the convolution layer with the RNN deep learning model, i.e., Convolution-Long short-term memory network as Convolution-LSTM and the bi-directional long short-term memory (BiLSTM) network as Convolution-BiLSTM. The experimental results demonstrate that, in terms of many metrics, i.e., network latency, packet loss ratio, network throughput, and overhead, our proposed methodologies perform better than the existing works, i.e., OSPF, FlowDCN, modified discrete PSO, ANN, CNN, and LSTM-based routing approaches. The proposed approach improves the network throughput by approximately 30% and 12% as compared to existing CNN and LSTM-based routing approaches, respectively.
Free
Deep Learning Sign Language Recognition System Based on Wi-Fi CSI
Scientific article
Many sensing gesture recognition systems based on Wi-Fi signals are introduced because of the commercial off-the-shelf Wi-Fi devices without any need for additional equipment. In this paper, a deep learning-based sign language recognition system is proposed. Wi-Fi CSI amplitude and phase information is used as input to the proposed model. The proposed model uses three types of deep learning: CNN, LSTM, and ABLSTM with a complete study of the impact of optimizers, the use of amplitude and phase of CSI, and preprocessing phase. Accuracy, F-score, Precision, and recall are used as performance metrics to evaluate the proposed model. The proposed model achieves 99.855%, 99.674%, 99.734%, and 93.84% average recognition accuracy for the lab, home, lab + home, and 5 different users in a lab environment, respectively. Experimental results show that the proposed model can effectively detect sign gestures in complex environments compared with some deep learning recognition models.
Free
Scientific article
Adversarial attacks can be extremely dangerous, particularly in scenarios where the precision of facial expression identification is of utmost importance. Hiring adversarial training methods proves effective in mitigating these threats. Although effective, this technique requires large computing resources. This study aims to strengthen deep learning model resilience against adversarial attacks while optimizing performance and resource efficiency. Our proposed method uses adversarial training techniques to create adversarial examples, which are permanently stored as a separate dataset. This strategy helps the model learn and enhances its resilience to adversarial attacks. This study also evaluates models by subjecting them to adversarial attacks, such as the One Pixel Attack and the Fast Gradient Sign Method, to identify any potential vulnerabilities. Moreover, we use two different model architectures to see how well they are protected against adversarial attacks. It compared their performances to determine the best model for making systems more resistant while still maintaining good performance. The findings show that the combination of the proposed adversarial training technique and an efficient model architecture outcome in increased resistance to adversarial attacks. This also improves the reliability of the model and saves more resources for computation. This is evidenced by the high accuracy results achieved at 98.81% accuracy on the CK+ datasets. The adversarial training technique proposed in this study offers an efficient alternative to overcome the limitations of computational resources. This fortifies the model against adversarial attacks, resulting in significant increases in model resilience without loss of performance.
Free
Deep Learning in Character Recognition Considering Pattern Invariance Constraints
Scientific article
Character recognition is a field of machine learning that has been under research for several decades. The particular success of neural networks in pattern recognition and therefore character recognition is laudable. Research has also long shown that a single hidden layer network has the capability to approximate any function; while, the problems associated with training deep networks therefore led to little attention given to it. Recently, the breakthrough in training deep networks through various pre-training schemes have led to the resurgence and massive interest in them, significantly outperforming shallow networks in several pattern recognition contests; moreover the more elaborate distributed representation of knowledge present in the different hidden layers concords with findings on the biological visual cortex. This research work reviews some of the most successful pre-training approaches to initializing deep networks such as stacked auto encoders, and deep belief networks based on achieved error rates. More importantly, this research also parallels investigating the performance of deep networks on some common problems associated with pattern recognition systems such as translational invariance, rotational invariance, scale mismatch, and noise. To achieve this, Yoruba vowel characters databases have been used in this research.
Free
Defect Analysis Using Artificial Neural Network
Scientific article
This paper deals with detection of defects in the manufactured ceramic tiles to ensure high density quality. The problem is concerned with the automatic inspection of ceramic tiles using Artificial Neural Network (ANN). The performance of the technique has been evaluated theoretically and experimentally on samples. Architecture of the system involves binary matrix processing and utilization of Artificial Neural Network (ANN) to detect defects. The above automatic inspection procedures have been implemented and tested on company floor tiles. The results obtained confirmed the efficiency of the methodology in defect detection in raw tile and its relevance as a promising approach on matrix, as well as included in quality control and inspection programs.
Free
Defuzzification Index for Ranking of Fuzzy Numbers on the Basis of Geometric Mean
Scientific article
The importance of fuzzy numbers to express uncertainty in certain applications, concerned with decision making, is observed in a large number of problems of different kinds. In Decision making problems, the best of available alternatives is chosen to the possible extent. In the process of ordering the alternatives, ranking of fuzzy numbers plays a key role. A large volume of ranking methods, based on different features, have been available in this domain. Owing to the complicated nature of fuzzy numbers, the so far introduced methods suffered setbacks or posed difficulties or showed drawbacks in one context or other. In addition, some methods are lengthy and complicated to apply on concerned problems. In this article, a new ranking procedure based on defuzzification, stemmed from the concepts of geometric mean and height of a fuzzy number, is proposed. Finally, numerical comparisons are made with other existing procedures for testing and validation of proposed method with the support of some standard numerical examples.
Free
Delay Computation Using Fuzzy Logic Approach
Scientific article
The paper presents practical application of fuzzy sets and system theory in predicting delay, with reasonable accuracy, a wide range of factors pertaining to construction projects. In this paper we shall use fuzzy logic to predict delays on account of Delayed supplies and Labor shortage. It is observed that the project scheduling software use either deterministic method or probabilistic method for computation of schedule durations, delays, lags and other parameters. In other words, these methods use only quantitative inputs leaving-out the qualitative aspects associated with individual activity of work. The qualitative aspect viz., the expertise of the mason or the lack of experience can have a significant impact on the assessed duration. Such qualitative aspects do not find adequate representation in the Project Scheduling software. A realistic project is considered for which a PERT chart has been prepared using showing all the major activities in reasonable detail. This project has been periodically updated until its completion. It is observed that some of the activities are delayed due to extraneous factors resulting in the overall delay of the project. The software has the capability to calculate the overall delay through CPM (Critical Path Method) when each of the activity-delays is reported. We shall now demonstrate that by using fuzzy logic, these delays could have been predicted well in advance.
Free
Density based initialization method for k-means clustering algorithm
Scientific article
Data clustering is a basic technique to show the structure of a data set. K-means clustering is a widely acceptable method of data clustering, which follow a partitioned approach for dividing the given data set into non-overlapping groups. Unfortunately, it has the pitfall of randomly choosing the initial cluster centers. Due to its gradient nature, this algorithm is highly sensitive to the initial seed value. In this paper, we propose a kernel density-based method to compute an initial seed value for the k-means algorithm. The idea is to select an initial point from the denser region because they truly reflect the property of the overall data set. Subsequently, we are avoiding the selection of outliers as an initial seed value. We have verified the proposed method on real data sets with the help of different internal and external validity measures. The experimental analysis illustrates that the proposed method has better performance over the k-means, k-means++ algorithm, and other recent initialization methods.
Free
Depth-guided Hybrid Attention Swin Transformer for Physics-guided Self-supervised Image Dehazing
Scientific article
Image dehazing is a critical preprocessing step in computer vision, enhancing visibility in degraded conditions. Conventional supervised methods often struggle with generalization and computational efficiency. This paper introduces a self-supervised image dehazing framework leveraging a depth-guided Swin Transformer with hybrid attention. The proposed hybrid attention explicitly integrates CNN-style channel and spatial attention with Swin Transformer window-based self-attention, enabling simultaneous local feature recalibration and global context aggregation. By integrating a pre-trained monocular depth estimation model and a Swin Transformer architecture with shifted window attention, our method efficiently models global context and preserves fine details. Here, depth is used as a relative structural prior rather than a metric quantity, enabling robust guidance without requiring haze-invariant depth estimation. Experimental results on synthetic and real-world benchmarks demonstrate superior performance, with a PSNR of 23.01 dB and SSIM of 0.879 on the RESIDE SOTS-indoor dataset, outperforming classical physics-based dehazing (DCP) and recent self-supervised approaches such as SLAD, achieving a PSNR gain of 2.52 dB over SLAD and 6.39 dB over DCP. Our approach also significantly improves object detection accuracy by 0.15 mAP@0.5 (+32.6%) under hazy conditions, and achieves near real-time inference (≈35 FPS at 256x256 resolution on a single GPU), confirming the practical utility of depth-guided features. Here, we show that our method achieves an SSIM of 0.879 on SOTS-Indoor, indicating strong structural and color fidelity for a self-supervised dehazing framework.
Free
Scientific article
The amount of cane fiber carried by cane carrier varies due to non-uniformity of cane supply. The continuous variation of cane in chute during the cane juice extraction inversely affects the cane juice extraction efficiency of mill. This paper deals with the development and performance comparison of conventional controller and fuzzy controller to maintain the cane level in chute during juice extraction. The conventional controller is developed using VHDL language and simulated by using Xilinx ISE 14.5 version. The fuzzy controller is developed and simulated by using fuzzy toolbox of ‘MATLAB® version 7.11.0.584 (R2020b).
Free
