International Journal of Computer Applications in Technology (23 papers in press)

Regular Issues

• Lightweight crop pest identification algorithm under natural background  
  by Dong Benzhi, Wang Yaqi, Xu Dali 
  Abstract: Aiming at the problem of poor detection effect and low recognition accuracy of small target insects under the background of complex natural environment, proposes an improved Yolo v5s insect recognition algorithm. The channel attention mechanism is embedded in the backbone network. The adaptive spatial feature fusion (ASFF) structure is introduced in the PANet part, and dynamic weight parameters is used to assign different weights to feature maps of different scales, Finally, we change the loss function and non-maximum suppression strategy to improve the accuracy of bounding box positioning and the speed of regression. Experimental results show that the improved algorithm has a final average accuracy (mAP@0.5) of 97.8% in the D0 dataset and an average detection speed of 13.66 ms per image, which is more suitable for deployment in mobile and embedded devices to achieve real-time detection.
  Keywords: insect recognition; lightweight convolutional neural network; channel attention mechanism; adaptive spatial feature fusion.
  
  
• Enhancing the accuracy of transformer-based embeddings for sentiment analysis in social big data  
  by Wiem Zemzem, Moncef Tagina 
  Abstract: Social media have opened a venue for online users to post and share their opinions in different life aspects, which leads to big data. As a result, sentiment analysis has become a fast-growing field of research in natural language processing (NLP) owing to its central role in analysing the public's opinion in many areas, including advertising, business, and marketing. This study proposes a transformer-based approach, which integrates contextualized words with Part-Of-Speech (POS) embedding. Then, the enhanced word vector is forwarded to a hybrid deep learning architecture combining a Convolutional Neural Network (CNN) and a Bidirectional Long Short Term Memory (BiLSTM) to discover the post's sentiment. Extensive experiments on four review datasets from diverse domains demonstrate that the proposed method outperforms other machine learning approaches in terms of accuracy.
  Keywords: deep learning; sentiment analysis; word embedding; big data; natural language processing.
  
  
• Unsupervised machine learning schemes for cooperative spectrum sensing in cognitive radio  
  by RAJENDRA YELALWAR, YERRAM RAVINDER 
  Abstract: The major challenge in the development of recent wireless technology is spectrum scarcity which is addressed by introducing the Cognitive Radio (CR) technique. In CR, spectrum sensing is the most critical task that senses the surrounding environment to detect the presence of a primary User (PU) in the target spectrum. This paper proposes the machine learning (ML) enabled Cooperative Spectrum Sensing (CSS) approaches where the application of clustering algorithms for the eigenvalue based CSS under different fading channel conditions is explored. The sensing performance is analysed with different PUs, signal features, Signal to Noise Ratio (SNR) values, and channel conditions. Secondly, this work proposes the novel clustering based CSS framework for Non-orthogonal Multiple Access (NOMA) signal detection. The system performance is measured in terms of sensing accuracy and Receiver Operating Curve (ROC). The simulation results ensure the effectiveness of the proposed clustering based CSS framework compared to the existing work in terms of improved accuracy which is observed to be 92.5% for K means clustering based CSS framework for NOMA
  Keywords: spectrum sensing; machine learning; K-means; K-medoids; agglomerative; NOMA.
  
  
• FPGA implementation and Multisim simulation of a new four-dimensional two-scroll hyperchaotic system with coexisting attractors  
  by Sundarapandian Vaidyanathan, Esteban Tlelo-Cuautle, Khaled Benkouider, Aceng Sambas, Ciro Fabian Bermudez-Marquez, Samy Abdelwahab Safaan 
  Abstract: Field-programmable gate array (FPGA) design of a new four-dimensional two-scroll hyperchaotic system is investigated in this work. A detailed system modelling of the new system with a hyperchaotic attractor begins this work with phase plots, which is followed by a bifurcation study of the new system. Special dynamic properties such as multistability and symmetry are also investigated for the new system. Using Multisim software, a circuit model is designed and simulated for the new hyperchaotic system. FPGA design and Multisim simulation of the new system enable practical applications in science and engineering. The implementation of the FPGA design in this work is carried out by applying two numerical schemes, viz. Forward Euler and Trapezoidal methods. Experimental attractors observed in the oscilloscope show good match with the Matlab signal plots.The FPGA hardware resources are detailed for both numerical methods.
  Keywords: hyperchaos; bifurcation; symmetry; phase plots; hyperchaotic system;rnparameters; stability; multistability; circuit model; FPGA implementation.
  
  
• An abstraction-based approach to eliciting wisdom in intangible cultural heritage utilisation  
  by Yu Wang, Sicong Ma, Lin Zou, Hongji Yang 
  Abstract: Intangible cultural heritage utilisation action involves a variety of knowledge, such as culture, history, politics and suchlike. However, intangible cultural heritage is arduous to meet the customers requirement in current situation. A challenge for intangible culture heritage barriers is to provide 'wise' products to satisfy the customers requirements. In order to achieve this target, this paper presents a novel system based on abstraction techniques, aiming to search or generate wisdom in the intangible cultural heritage domain. It covers three phases, i.e., Abstracting Data/Information/Knowledge, Searching Metadata and Ranking the wisdom results. Abstraction techniques aim to analyse the characteristics of customers requirements and construct the levels of abstraction. Cuckoo Search is used for searching relevant metadata and potential relevant meta data to express the wisdoms. Furthermore, wisdom metrics are built to rank the wise products.
  Keywords: intangible cultural heritage; abstraction techniques and wisdom; creative computing.
  
  
• Improving hybrid-layer convolutional neural network system for lung cancer nodule classification using enhanced weight optimisation algorithm  
  by Vikul Pawar, P. Premchand 
  Abstract: In recent times, lung cancer is evolving as a highly life-threatening disease for human beings. According to the WHO, lung cancer disease is the second largest cause of deaths as compared to all other types of cancer. The prevailing available technology is striving to get more exposure in the field of medical science using Computer Assisted Diagnosis (CAD), where image processing is playing a crucial role for detecting the cancerous nodules in computer tomographic images. Augmenting the machine learning techniques with image processing algorithms is becoming a more comprehensive examination of cancer disease in proposed CAD systems. This paper is describes a heuristic approach for lung cancer nodule detection, and the proposed model predominantly consists of the following tasks, which are image enhancement, segmenting ROI (Region of Interest), features extraction, and nodule classification. In pre-processing, primarily the Adaptive Median Filter (AMF) filtering method is applied to eliminate the speckle noise from input CT images of Lung Image Database Consortium (LIDC) and Image Database Resource Initiative (IDRI): in the LIDC-IDRI dataset, the quality of input image is improved by applying Histogram Equalization (HE) technique with Contrast-Limited Adaptive (CLA) approach. Secondly, in the successive stage the Improved Level-Set (ILS) algorithm is used to segment the ROI. Furthermore, the third step of the projected work is applied to extract the definite learnable texture features and statistical features from the segmented ROI. The extracted features in the subsequent stage of classification are applied to Hybrid-Layer Convolutional Neural Network (HL-CNN) architecture to classify the lung cancer nodule as either benign or malignant. Principally this research is carried out by contributing to each stage of it, where the novel concept of the improved Hybrid-Layer Convolutional Neural Network (HL-CNN) is employed by optimising and selecting the optimal weight using the Enhanced Cat Swarm Optimisation (ECSO) algorithm. The experimental result of the proposed HL-CNN using the weight optimisation algorithm ECSO is achieved an accuracy of 93%, which is comparatively efficient with respect to existing models such as DBN, SVM, CNN, WOA, MFO, and CSO. Moreover, the proposed model conclusively gives a decision on the detected nodule as either benign or malignant.
  Keywords: Computer Assisted Diagnosis (CAD); Computer Vision; Cancer Diagnosis; Image Classification; Image Enhancement; Image Segmentation; Feature Extraction.
  
  
• Prediction model for total amount of coke oven gas generation based on FCM-RBF  
  by Lili Feng, Jun Peng, Zhaojun Huang 
  Abstract: The rational use of Coke Oven Gas (COG) is of great significance to improve the economic efficiency of enterprises. In this paper, a COG generation prediction model based on fuzzy C-mean clustering (FCM) and radial basis function (RBF) neural network is proposed to address the problems such as the difficulty of accurate modelling of COG generation process and the difficulty of real-time flow prediction. Firstly, the coke oven production process is analysed and correlation analysis is used to select the influencing factors. Secondly, the FCM is used to classify the working conditions of the coke oven, and the appropriate number of working conditions is selected through experiments. Finally, the prediction models under different working conditions are established separately by using RBF. The experiments were carried out using actual industrial production data, and the experimental results showed that the model could provide guidance reference for the dispatchers.
  Keywords: coking oven process; fuzzy C-means clustering; prediction model; radial basis function neural network.
  
  
• Efficient load balancing in cloud computing using HHO improved by differential perturbed velocity and TEO  
  by Uttam Kumar Jena, Manas Ranjan Kabat, Pradipta Kumar Das 
  Abstract: Load balancing is one of the primary aspects of cloud computing to avoid situations of being overloaded or underloaded in the node. This paper aims to carry out the dynamic load balancing of non-determent independent tasks in the cloud network and resolved through the hybridisation of an improved version of the Harris Hawk Optimisation Algorithm (HHO) improved by differential perturbed velocity and Thermal Exchange optimization (TEO). The main motivation of hybridising is to intensify the diversification ability of the device through the load balance with the VMs, in order to optimise different matrices and enhance the convergence speed. The strength of the algorithm has been authenticated by relating the outcome gained from simulation and real platform processes with the surviving load balancing. The conclusions drawn from the simulation and comparison results illustrate that the projected procedure is outstripping its opponent in the manner of different matrices.
  Keywords: load balancing; throughput time; cloud network; optimisation.
  
  
• Group delay-based minimum variance distortion-less response cepstral features for speaker identification in whispered speech  
  by Vijay Sardar, Manisha L. Jadhav, Saurabh H. Deshmukh, Makarand M. Jadhav 
  Abstract: The whispering voice shows a wide difference in characteristics compared to the neutral voice. It makes identification of a person from the whispered sound difficult. The group delay function (GDF) in its spectral form considers the phase information in the short-time FT phase function, which is otherwise ignored in traditional front-end processing. A minimum variance distortion-less response (MVDR) based on smoothing on the denominator of group delay that enhances speech quality and intelligibility is proposed in this paper. The experiment uses MVDR spectral coefficient features with a multi-class Support Vector Machine (SVM) for classification. The proposed method reported an improvement of 2.41% over the baseline system using the CHAINs database and SVM classifier. The five-fold cross-validation is exercised for accuracy and speaker error rate to verify the consistency of the results. The proposed system is also evaluated for false positives and precision and reported the enhancement compared to the baseline system.
  Keywords: whispered speech; group delay function; MVDR; support vector machine; MFCC.
  
  
• Providing an open framework to facilitate tax fraud detection  
  by Jean Prolhac, Christophe Gaie 
  Abstract: In the present article, the authors propose a novel framework to optimize tax fraud detection. The framework is based on the construction of four successive layers: Modelling, Datasets, Algorithms, and Interfaces. The fraud detection model relies on four parameters that enable the computation of four fraud indicators. The datasets are built using five open data sources provided by the French Statistics Department. Then, fraud detection is performed using different neural network and random forest algorithms whose performances are discussed. Thereafter a novel approach is proposed as tax files are assigned to auditors according to their experience and skills. Finally, the interfaces developed during the project are described and offer a simple manner to benefit from the proposed framework. As the framework is shared on a public repository, every researcher can now contribute to optimize fraud detection algorithms.
  Keywords: artificial intelligence; neural networks; e-government; fraud detection; open framework.
  DOI: 10.1504/IJCAT.2023.10055494
   
  
• An emotion-aware search engine for multimedia content based on deep learning algorithms  
  by Andrea Chiorrini, Claudia Diamantini, Alex Mircoli, Domenico Potena, Emanuele Storti 
  Abstract: Nowadays, large amounts of unstructured data are available online. Such data often contain users emotions and feelings about a variety of topics but their retrieval and selection on the basis of an emotional perspective are usually unfeasible through traditional search engines, which only rank Web content according to its relevance with respect to a given search keyword. For this reason, in the present work we introduce the architecture of a novel emotion-aware search engine that can return search results ranked on the basis of seven human emotions. Using this system, users can benefit from a more advanced semantic search that also takes into account emotions. The system uses emotion recognition algorithms based on deep learning to extract emotion vectors from texts, images and videos and then populates an emotional index to allow users to visualize results related to given emotions. We also discuss and evaluate different deep learning models for building emotional indexes from texts, images and videos.
  Keywords: emotion recognition; query answering; emotion-aware query answering; multimedial query answering; sentiment analysis; emotion analysis; emotion-aware search engine; deep learning; BERT; multimodal analysis.
  
  
• Time integration algorithm of velocity frequency-weighted root mean square for ship comfort index  
  by Jin Yan, Jian-bin Liao, Han-lin Li, Hong-liang Yu 
  Abstract: The imposition of excessive ship vibration induces corresponding motions and forces within the human body, possibly causing discomfort and degraded levels of comfort and health. The ISO criteria is designed to limit potential vibration to improve the crew comfort. Frequency weighting values are used to evaluate whole-body vibration in accordance with ISO standard. This study described the method using frequency method to compute the frequency weighted velocity values by using acceleration time signals firstly, the relationship between velocity and acceleration is deduced in detail. And then a time domain integrated method is proposed to calculate the frequency weighted velocity RMS, finally, the validity of the method is assessed on a real sea test experiment and compared with frequency domain method.
  Keywords: ship habitability evaluation; vibration assessment; time domain integration.
  DOI: 10.1504/IJCAT.2023.10056921
   
  
• Deep learning approach based hybrid fine-tuned Smith algorithm with Adam optimiser for multilingual opinion mining  
  by Aniket K. Shahade, K.H. Walse, V.M. Thakare 
  Abstract: A deep learning-based Hybrid Fine Tuned Smith Algorithm with Adam optimizer (HFS-AO) is introduced for multilingual opinion mining. Initially, data are collected using the web scraping algorithm to collect three different languages data: Marathi, Hindi and English. After the data extraction, the annotation process is suggested to label the collected data using the Zero-shot instance-weighting technique. Further, pre-process the data to remove unnecessary noises and symbols. After that, text vectorization is performed using Na
  Keywords: opinion mining; fine tuned Smith algorithm; Adam optimiser; annotation; pre-processing; stop-word removal; tokenisation; lemmatisation; stemming; POS tagging; naive Bayes vectorisation; Text Vectoriza.
  DOI: 10.1504/IJCAT.2023.10057079
   
  
• A summative survey on ECG and PPG signal analysis  
  by Tara Rasti, Quanmin Zhu, Keming Zhou, Aakash Soni, Shanshan Zhao, Duncan Sleeman 
  Abstract: The heart electrical activity can be measured with an Electrocardiogram (ECG), where electrodes are located on the skin. The second approach is the photoplethysmography (PPG) to measure the changes of the blood volume by the light absorbed/reflected from tissues. ECG has higher accuracy than PPG in measurements, but requires a closed-loop circuit to complete a measurement. Although they have the merit of non-invasive measurements, ECG and PPG signals are non-stationary, very sensitive to noise, and highly stochastically variable. Therefore, they need to be pre-processed to denoises. Generally, the analysis of ECG and PPG signals involves pre-processing and feature extraction/selection/classification. Furthermore, appropriate configuration of the ECG/PPG signal database and performance specification, such as accuracy, specificity, sensitivity, etc., have significance in the signal analysis. This paper presents an updated literature survey on ECG/PPG signal analysis for future study reference. The review focuses on the electrical signal processing instead of medical interpretation.
  Keywords: electrocardiogram; photoplethysmography; signal pre-processing; feature extraction; feature selection; classification; emotion assessment; arrhythmia detection.
  
  
• Design and implementation of a low-cost and multi-parameter indoor air quality detector based on IoT  
  by Liang Zhao, Yujun Yao, Shuai Huang, Jieru Xu 
  Abstract: Nowadays, severe air pollution has become a leading environmental risk. Detection of indoor air pollutants may be more crucial than outdoor, because people spend nearly 90% of their time indoors. This paper proposes a low-cost and multi-parameter indoor air quality detector, able to measure seven significant indoor air parameters. Firstly, based on IoT concept, the system architecture of the proposed detector is introduced. Subsequently, by adopting modular method, the hardware circuit of various air parameter acquisition modules is designed. Then, based on I2C and UART protocol, the corresponding software programs for each module are developed. Moreover, the proposed system has been deployed in an office room of China Three Gorges University (CTGU) and stably operated for nearly a year. The test results show that the proposed system can accurately sense and reliably upload multiple air indexes including temperature, humidity, PM2.5, PM10, CO2, HCHO and VOC, which is suitable for IoT applications like smart home and smart building.
  Keywords: indoor air quality; low-cost; multi-parameter; detector; Internet of Things.
  DOI: 10.1504/IJCAT.2023.10058626
   
  
• Evaluation of VDT visual fatigue based on real-time signal processing of finger Photoplethysmography pulse wave  
  by Yetai Wang, Lijuan Chou, Yiquan Sun, Yongxin Chou 
  Abstract: In this paper, we extract parameters related to visual fatigue from the biological signal during VDT operation using finger photoplethysmography (PPG) pulse wave. The analysis results has shown that the peak-to-trough amplitudes and magnitude spectrum peaks are all significantly decreased after visual fatigue experiments (p<0.001) and the classification accuracy of fatigue states reach 94.59%. This provides a simple convenient and objective detection method for diagnosis and evaluation VDT visual fatigue. The low computational complexity and good performance on fatigue detection have great potential to be implemented in real-time for the primary care and in time adopting measures to effectively prevent the occurrence and the development of the visual fatigue. So the technique of PPG could be used in a wider range of clinical measurements as a measure of visual fatigue state in real-time monitoring.
  Keywords: visual display terminal; visual fatigue; photoplethysmography; peak-to-trough amplitude; magnitude spectrum peaks.
  DOI: 10.1504/IJCAT.2023.10058627
   
  
• The compound iterative algorithm for rational models based on the coyote optimisation algorithm  
  by Fei Xv, Jing Chen, Xia Yin 
  Abstract: This article proposes a coyote optimisation compound iterative algorithm (CO-CIA) for rational models. Particularly, the parameters in the numerator and denominator of rational models make the derivative equation hard to solve. To deal with this problem, the coyote optimisation algorithm (COA) is applied to estimate the parameters in the denominator. Compared with the bias compensation based least squares (BCLS) algorithm and the particle swarm optimisation compound iterative algorithm (PSO-CIA), the proposed method has higher accuracy and faster convergence rates. Finally, a simulation example is used to verify the effectiveness of the proposed algorithm.
  Keywords: coyote optimisation algorithm; recursive least squares; rational model; parameter estimation; compound iterative algorithm.
  DOI: 10.1504/IJCAT.2023.10058715
   
  
• Bio-inspired method for segmenting the optic disc and macula in retinal images  
  by Ahmad Taher Azar 
  Abstract: In this paper, ant colony system (ACS) based segmentation method is proposed and its features is used for segmenting the optic disc and macula in the retinal images. The suggested ACS-based segmentation approach employs intensity (grey-level) and colour as distinct characteristics to localize and segment the optic disc and macula in the retinal images. The evaluation and assessment of the performance of this nature-inspired approach for automatically separating blood vessels in retinal images by computer on standard retinal image databases revealed that it obtained the best separation of pixels only that belong to a blood vessel above all previous findings. The degree of accuracy of the proposed methodology approaches the best levels of accuracy achieved by techniques using supervised pixel classification algorithms.
  Keywords: ant colony system; bio-inspired algorithms; classification; retinal images; computer-aided diagnosis.
  DOI: 10.1504/IJCAT.2023.10058751
   
  
• Improving exploration in deep reinforcement learning for stock trading  
  by Wiem Zemzem, Moncef Tagina 
  Abstract: Deep reinforcement learning techniques have become quite widespread over the last decades. One challenge is the Exploration-Exploitation Dilemma. Although many exploration techniques for single-agent and multi-agent deep reinforcement learning are proposed and have shown promising results in various domains, their value was not yet been demonstrated in the financial markets. In this paper, we will apply the NoisyNet-DQN method, which was previously tested and brought promising results in Atari games, to the stock trading problem. The trained reinforcement learning agent is employed to trade the S&P500 ETF (SPY) dataset. Findings show that this approach can encounter the best trading action to choose at a specific moment and outperforms the classical DQN method.
  Keywords: deep reinforcement learning; NoisyNet-DQN; exploration; stock trading.
  DOI: 10.1504/IJCAT.2023.10058882
   
  
• Cryptographic applications of a new hyperchaotic system with a hidden attractor  
  by Sundarapandian Vaidyanathan, Irene Moroz, Aceng Sambas, Bassem Abd-El-Atty, Ahmed A. Abd El-Latif 
  Abstract: This work puts forth a new 4-D hyperchaotic 2-scroll dynamics with a hidden attractor. The proposed nonlinear system has three quadratic terms and it is established that there is no fixed point for the proposed dynamics for all values of the system constants. Bifurcation analysis for the proposed system is investigated in detail. Next, we invoke Multisim for the development of an electronic circuit for the proposed hyperchaotic dynamics. Finally, we present some cryptographic applications of the proposed hyperchaotic dynamics. A new algorithm for the construction of substitution boxes (S-boxes) using the presented hyperchaotic system is proposed, in which the performance of the generated S-box demonstrates its efficiency and reliability for designing modern block-cipher cryptosystems. The proposed hyperchaotic system is utilised for generating pseudo-random number generators (PRNGs), in which the results of NIST SP 800-22 tests demonstrate the randomness characteristics of the generated PRNG sequences.
  Keywords: hyperchaos; hyperchaotic system; chaos; bifurcations; circuit design; Multisim; cryptography; cryptosystem.
  DOI: 10.1504/IJCAT.2023.10058883
   
  
• Link prediction analysis based on Node2vec embedding technique  
  by Salam Jayachitra Devi, Buddha Singh 
  Abstract: Link prediction analysis can aid in better understanding of complex networks. Traditional link prediction algorithms suffer from high computational cost with excessive memory requirements for analysis of huge size network. Therefore, embedding techniques are used in network analysis with less memory utilization by converting high dimensional sparse networks into low dimensional vector preserving the structural properties of network. In embedding technique every node’s property present in a network are embedded into a vector of smaller dimension. Later, node’s similarity of the original complex network with irregular spaces can be determined from the embedded node. Hence, this paper contributes on the analysis of link prediction using Node2Vec embedding technique based on Random walk algorithm. Several machine learning models were also used for analysing the embedding technique. The Node2Vec embedding technique used various edge embedding operators such as Hadamard, Concatenation, Average, Weighted L1 and Weighted L2. The analysis of this embedding technique is done on real world network datasets obtained from various domain.
  Keywords: embedding; link prediction; random walk; Node2Vec; natural language processing.
  DOI: 10.1504/IJCAT.2023.10058967
   
  
• Computer-aided crop disease classification system using colour and texture features  
  by Megha Agarwal 
  Abstract: Reliable and fast plant disease recognition system is very important for both human health and the economy of the world. It is required to design an automatic system to predict the crop diseases and estimate its production accordingly. Several factors are causing damages to the crop hence in this paper, features are extracted through colour and texture properties of tomato plant leaves, and supervised classification is performed. Images are decomposed into different frequency sub-bands using difference of Gaussian filters and features are extracted through local ternary co-occurrence patterns. Colour properties are added through hue and saturation components. This comprehensive feature is used to perform classification on PlantVillage dataset. Performance is evaluated using machine learning classifiers. The proposed hand-crafted feature is able to perform superior to the deep learning models as well as state-of-the-art methods in terms of classification accuracy and AUC.
  Keywords: disease classification; plant disease; PlantVillage; texture feature.
  DOI: 10.1504/IJCAT.2023.10058968
   
  
• Modelling and simulation of portable solar Scheffler reflector water heater using soft computing techniques  
  by Mangesh Phate, Shraddha Toney, Vikas Phate 
  Abstract: Experimental analysis of the solar Scheffler reflector was carried out for the performance improvement in the Indian climate. A dish of 2 7 m2 size has been designed for the study. The maximum capacity of 15 lit received was used for the analysis, which serves as a hot water storage system. The thermal efficiency was estimated based on the reflector's incident radiation shape, operation time, dish area, altitude angle, the initial temperature of the water, tilting angle of the reflector. The response surface method and artificial neural network were employed to analyse the reflector's performance. The outcomes showed that a good agreement was observed between the model prediction and the measured efficiency. This work made an effort to predict the Scheffler reflector's thermal efficiency for water heating. The statistical parameters calculated between actual or experimental efficiency with the model predicted efficiency.
  Keywords: artificial neural network; solar Scheffler reflector; response surface method; solar energy; efficiency; hot water system.
  DOI: 10.1504/IJCAT.2023.10059367