Search results

The paper aims to tackle the classification of Remote Sensing Images (RSIs), which presents a significant challenge for computer algorithms due to the inherent characteristics of clustered ground objects and noisy backgrounds. Recent research typically leverages larger volume models to achieve advanced performance. However, the operating environments of remote sensing commonly cannot provide unconstrained computational and storage resources. It requires lightweight algorithms with exceptional generalization capabilities.

This study introduces an efficient knowledge distillation (KD) method to build a lightweight yet precise convolutional neural network (CNN) classifier. This method also aims to substantially decrease the training time expenses commonly linked with traditional KD techniques. This approach entails extensive alterations to both the model training framework and the distillation process, each tailored to the unique characteristics of RSIs. In particular, this study establishes a robust ensemble teacher by independently training two CNN models using a customized, efficient training algorithm. Following this, this study modifies a KD loss function to mitigate the suppression of non-target category predictions, which are essential for capturing the inter- and intra-similarity of RSIs.

This study validated the student model, termed KD-enhanced network (KDE-Net), obtained through the KD process on three benchmark RSI data sets. The KDE-Net surpasses 42 other state-of-the-art methods in the literature published from 2020 to 2023. Compared to the top-ranked method’s performance on the challenging NWPU45 data set, KDE-Net demonstrated a noticeable 0.4% increase in overall accuracy with a significant 88% reduction in parameters. Meanwhile, this study’s reformed KD framework significantly enhances the knowledge transfer speed by at least three times.

This study illustrates that the logit-based KD technique can effectively develop lightweight CNN classifiers for RSI classification without substantial sacrifices in computation and storage costs. Compared to neural architecture search or other methods aiming to provide lightweight solutions, this study’s KDE-Net, based on the inherent characteristics of RSIs, is currently more efficient in constructing accurate yet lightweight classifiers for RSI classification.

This research paper aims to analyze the scientific and grey literature on Quality 4.0 and zero-defect manufacturing (ZDM) frameworks to develop an integrated quality 4.0 framework (IQ4.0F) for quality improvement (QI) based on Six Sigma and machine learning (ML) techniques towards ZDM. The IQ4.0F aims to contribute to the advancement of defect prediction approaches in diverse manufacturing processes. Furthermore, the work enables a comprehensive analysis of process variables influencing product quality with emphasis on the use of supervised and unsupervised ML techniques in Six Sigma’s DMAIC (Define, Measure, Analyze, Improve and Control) cycle stage of “Analyze.”

The research methodology employed a systematic literature review (SLR) based on PRISMA guidelines to develop the integrated framework, followed by a real industrial case study set in the automotive industry to fulfill the objectives of verifying and validating the proposed IQ4.0F with primary data.

This research work demonstrates the value of a “stepwise framework” to facilitate a shift from conventional quality management systems (QMSs) to QMSs 4.0. It uses the IDEF0 modeling methodology and Six Sigma’s DMAIC cycle to structure the steps to be followed to adopt the Quality 4.0 paradigm for QI. It also proves the worth of integrating Six Sigma and ML techniques into the “Analyze” stage of the DMAIC cycle for improving defect prediction in manufacturing processes and supporting problem-solving activities for quality managers.

This research paper introduces a first-of-its-kind Quality 4.0 framework – the IQ4.0F. Each step of the IQ4.0F was verified and validated in an original industrial case study set in the automotive industry. It is the first Quality 4.0 framework, according to the SLR conducted, to utilize the principal component analysis technique as a substitute for “Screening Design” in the Design of Experiments phase and K-means clustering technique for multivariable analysis, identifying process parameters that significantly impact product quality. The proposed IQ4.0F not only empowers decision-makers with the knowledge to launch a Quality 4.0 initiative but also provides quality managers with a systematic problem-solving methodology for quality improvement.

Although numerous signal modalities are available for emotion recognition, audio and visual modalities are the most common and predominant forms for human beings to express their emotional states in daily communication. Therefore, how to achieve automatic and accurate audiovisual emotion recognition is significantly important for developing engaging and empathetic human–computer interaction environment. However, two major challenges exist in the field of audiovisual emotion recognition: (1) how to effectively capture representations of each single modality and eliminate redundant features and (2) how to efficiently integrate information from these two modalities to generate discriminative representations.

A novel key-frame extraction-based attention fusion network (KE-AFN) is proposed for audiovisual emotion recognition. KE-AFN attempts to integrate key-frame extraction with multimodal interaction and fusion to enhance audiovisual representations and reduce redundant computation, filling the research gaps of existing approaches. Specifically, the local maximum–based content analysis is designed to extract key-frames from videos for the purpose of eliminating data redundancy. Two modules, including “Multi-head Attention-based Intra-modality Interaction Module” and “Multi-head Attention-based Cross-modality Interaction Module”, are proposed to mine and capture intra- and cross-modality interactions for further reducing data redundancy and producing more powerful multimodal representations.

Extensive experiments on two benchmark datasets (i.e. RAVDESS and CMU-MOSEI) demonstrate the effectiveness and rationality of KE-AFN. Specifically, (1) KE-AFN is superior to state-of-the-art baselines for audiovisual emotion recognition. (2) Exploring the supplementary and complementary information of different modalities can provide more emotional clues for better emotion recognition. (3) The proposed key-frame extraction strategy can enhance the performance by more than 2.79 per cent on accuracy. (4) Both exploring intra- and cross-modality interactions and employing attention-based audiovisual fusion can lead to better prediction performance.

The proposed KE-AFN can support the development of engaging and empathetic human–computer interaction environment.

Bank creditworthiness refers to the evaluation of a bank’s ability to meet its financial obligations. It is an assessment of the bank’s financial health, stability and capacity to manage risks. This paper aims to investigate the credit rating patterns that are crucial for assessing creditworthiness of the Islamic banks, thereby evaluating the stability of their industry.

Three distinct machine learning algorithms are exploited and evaluated for the desired objective. This research initially uses the decision tree machine learning algorithm as a base learner conducting an in-depth comparison with the ensemble decision tree and Random Forest. Subsequently, the Apriori algorithm is deployed to uncover the most significant attributes impacting a bank’s credit rating. To appraise the previously elucidated models, a ten-fold cross-validation method is applied. This method involves segmenting the data sets into ten folds, with nine used for training and one for testing alternatively ten times changeable. This approach aims to mitigate any potential biases that could arise during the learning and training phases. Following this process, the accuracy is assessed and depicted in a confusion matrix as outlined in the methodology section.

The findings of this investigation reveal that the Random Forest machine learning algorithm superperforms others, achieving an impressive 90.5% accuracy in predicting credit ratings. Notably, our research sheds light on the significance of the loan-to-deposit ratio as a primary attribute affecting credit rating predictions. Moreover, this study uncovers additional pivotal banking features that intensely impact the measurements under study. This paper’s findings provide evidence that the loan-to-deposit ratio looks to be the purest bank attribute that affects credit rating prediction. In addition, deposit-to-assets ratio and profit sharing investment account ratio criteria are found to be effective in credit rating prediction and the ownership structure criterion came to be viewed as one of the essential bank attributes in credit rating prediction.

These findings contribute significant evidence to the understanding of attributes that strongly influence credit rating predictions within the banking sector. This study uniquely contributes by uncovering patterns that have not been previously documented in the literature, broadening our understanding in this field.

The traditional failure mode and effect analysis (FMEA) has some limitations, such as the neglect of relevant historical data, subjective use of rating numbering and the less rationality and accuracy of the Risk Priority Number. The current study proposes a machine learning–enhanced FMEA (ML-FMEA) method based on a popular machine learning tool, Waikato environment for knowledge analysis (WEKA).

This work uses the collected FMEA historical data to predict the probability of component/product failure risk by machine learning based on different commonly used classifiers. To compare the correct classification rate of ML-FMEA based on different classifiers, the 10-fold cross-validation is employed. Moreover, the prediction error is estimated by repeated experiments with different random seeds under varying initialization settings. Finally, the case of the submersible pump in Bhattacharjee et al. (2020) is utilized to test the performance of the proposed method.

The results show that ML-FMEA, based on most of the commonly used classifiers, outperforms the Bhattacharjee model. For example, the ML-FMEA based on Random Committee improves the correct classification rate from 77.47 to 90.09 per cent and area under the curve of receiver operating characteristic curve (ROC) from 80.9 to 91.8 per cent, respectively.

The proposed method not only enables the decision-maker to use the historical failure data and predict the probability of the risk of failure but also may pave a new way for the application of machine learning techniques in FMEA.

Agriculture is the backbone of a country, contributing more than half of the sector of economy throughout the world. The need for precision agriculture is essential in evaluating the conditions of the crops with the aim of determining the proper selection of pesticides. The conventional method of pest detection fails to be stable and provides limited accuracy in the prediction. This paper aims to propose an automatic pest detection module for the accurate detection of pests using the hybrid optimization controlled deep learning model.

The paper proposes an advanced pest detection strategy based on deep learning strategy through wireless sensor network (WSN) in the agricultural fields. Initially, the WSN consisting of number of nodes and a sink are clustered as number of clusters. Each cluster comprises a cluster head (CH) and a number of nodes, where the CH involves in the transfer of data to the sink node of the WSN and the CH is selected using the fractional ant bee colony optimization (FABC) algorithm. The routing process is executed using the protruder optimization algorithm that helps in the transfer of image data to the sink node through the optimal CH. The sink node acts as the data aggregator and the collection of image data thus obtained acts as the input database to be processed to find the type of pest in the agricultural field. The image data is pre-processed to remove the artifacts present in the image and the pre-processed image is then subjected to feature extraction process, through which the significant local directional pattern, local binary pattern, local optimal-oriented pattern (LOOP) and local ternary pattern (LTP) features are extracted. The extracted features are then fed to the deep-convolutional neural network (CNN) in such a way to detect the type of pests in the agricultural field. The weights of the deep-CNN are tuned optimally using the proposed MFGHO optimization algorithm that is developed with the combined characteristics of navigating search agents and the swarming search agents.

The analysis using insect identification from habitus image Database based on the performance metrics, such as accuracy, specificity and sensitivity, reveals the effectiveness of the proposed MFGHO-based deep-CNN in detecting the pests in crops. The analysis proves that the proposed classifier using the FABC+protruder optimization-based data aggregation strategy obtains an accuracy of 94.3482%, sensitivity of 93.3247% and the specificity of 94.5263%, which is high as compared to the existing methods.

The proposed MFGHO optimization-based deep-CNN is used for the detection of pest in the crop fields to ensure the better selection of proper cost-effective pesticides for the crop fields in such a way to increase the production. The proposed MFGHO algorithm is developed with the integrated characteristic features of navigating search agents and the swarming search agents in such a way to facilitate the optimal tuning of the hyperparameters in the deep-CNN classifier for the detection of pests in the crop fields.

While the Chinese securities market is booming, the phenomenon of listed companies falling into financial distress is also emerging, which affects the operation and development of enterprises and also jeopardizes the interests of investors. Therefore, it is important to understand how to accurately and reasonably predict the financial distress of enterprises.

In the present study, ensemble feature selection (EFS) and improved stacking were used for financial distress prediction (FDP). Mutual information, analysis of variance (ANOVA), random forest (RF), genetic algorithms, and recursive feature elimination (RFE) were chosen for EFS to select features. Since there may be missing information when feeding the results of the base learner directly into the meta-learner, the features with high importance were fed into the meta-learner together. A screening layer was added to select the meta-learner with better performance. Finally, Optima hyperparameters were used for parameter tuning by the learners.

An empirical study was conducted with a sample of A-share listed companies in China. The F1-score of the model constructed using the features screened by EFS reached 84.55%, representing an improvement of 4.37% compared to the original features. To verify the effectiveness of improved stacking, benchmark model comparison experiments were conducted. Compared to the original stacking model, the accuracy of the improved stacking model was improved by 0.44%, and the F1-score was improved by 0.51%. In addition, the improved stacking model had the highest area under the curve (AUC) value (0.905) among all the compared models.

Compared to previous models, the proposed FDP model has better performance, thus bridging the research gap of feature selection. The present study provides new ideas for stacking improvement research and a reference for subsequent research in this field.

This study has mainly aimed to compare and contrast two completely different image processing algorithms that are very adaptive for detecting prostate cancer using wearable Internet of Things (IoT) devices. Cancer in these modern times is still considered as one of the most dreaded disease, which is continuously pestering the mankind over a past few decades. According to Indian Council of Medical Research, India alone registers about 11.5 lakh cancer related cases every year and closely up to 8 lakh people die with cancer related issues each year. Earlier the incidence of prostate cancer was commonly seen in men aged above 60 years, but a recent study has revealed that this type of cancer has been on rise even in men between the age groups of 35 and 60 years as well. These findings make it even more necessary to prioritize the research on diagnosing the prostate cancer at an early stage, so that the patients can be cured and can lead a normal life.

The research focuses on two types of feature extraction algorithms, namely, scale invariant feature transform (SIFT) and gray level co-occurrence matrix (GLCM) that are commonly used in medical image processing, in an attempt to discover and improve the gap present in the potential detection of prostate cancer in medical IoT. Later the results obtained by these two strategies are classified separately using a machine learning based classification model called multi-class support vector machine (SVM). Owing to the advantage of better tissue discrimination and contrast resolution, magnetic resonance imaging images have been considered for this study. The classification results obtained for both the SIFT as well as GLCM methods are then compared to check, which feature extraction strategy provides the most accurate results for diagnosing the prostate cancer.

The potential of both the models has been evaluated in terms of three aspects, namely, accuracy, sensitivity and specificity. Each model’s result was checked against diversified ranges of training and test data set. It was found that the SIFT-multiclass SVM model achieved a highest performance rate of 99.9451% accuracy, 100% sensitivity and 99% specificity at 40:60 ratio of the training and testing data set.

The SIFT-multi SVM versus GLCM-multi SVM based comparison has been introduced for the first time to perceive the best model to be used for the accurate diagnosis of prostate cancer. The performance of the classification for each of the feature extraction strategies is enumerated in terms of accuracy, sensitivity and specificity.

In order to estimate the value of semi-automated subject indexing in operative library catalogues, the study aimed to investigate five different automated implementations of an open source software package on a large set of Swedish union catalogue metadata records, with Dewey Decimal Classification (DDC) as the target classification system. It also aimed to contribute to the body of research on aboutness and related challenges in automated subject indexing and evaluation.

On a sample of over 230,000 records with close to 12,000 distinct DDC classes, an open source tool Annif, developed by the National Library of Finland, was applied in the following implementations: lexical algorithm, support vector classifier, fastText, Omikuji Bonsai and an ensemble approach combing the former four. A qualitative study involving two senior catalogue librarians and three students of library and information studies was also conducted to investigate the value and inter-rater agreement of automatically assigned classes, on a sample of 60 records.

The best results were achieved using the ensemble approach that achieved 66.82% accuracy on the three-digit DDC classification task. The qualitative study confirmed earlier studies reporting low inter-rater agreement but also pointed to the potential value of automatically assigned classes as additional access points in information retrieval.

The paper presents an extensive study of automated classification in an operative library catalogue, accompanied by a qualitative study of automated classes. It demonstrates the value of applying semi-automated indexing in operative information retrieval systems.

A novel method has been proposed to reduce the false alarm rate of arrhythmia patients regarding life-threatening conditions in the intensive care unit. In this purpose, the atrial blood pressure, photoplethysmogram (PLETH), electrocardiogram (ECG) and respiratory (RESP) signals are considered as input signals.

Three machine learning approaches feed-forward artificial neural network (ANN), ensemble learning method and k-nearest neighbors searching methods are used to detect the false alarm. The proposed method has been implemented using Arduino and MATLAB/SIMULINK for real-time ICU-arrhythmia patients' monitoring data.

The proposed method detects the false alarm with an accuracy of 99.4 per cent during asystole, 100 per cent during ventricular flutter, 98.5 per cent during ventricular tachycardia, 99.6 per cent during bradycardia and 100 per cent during tachycardia. The proposed framework is adaptive in many scenarios, easy to implement, computationally friendly and highly accurate and robust with overfitting issue.

As ECG signals consisting with PQRST wave, any deviation from the normal pattern may signify some alarming conditions. These deviations can be utilized as input to classifiers for the detection of false alarms; hence, there is no need for other feature extraction techniques. Feed-forward ANN with the Lavenberg–Marquardt algorithm has shown higher rate of convergence than other neural network algorithms which helps provide better accuracy with no overfitting.