Journal of Supercomputing最新文献

Sentiment analysis involves extricating and interpreting people's views, feelings, beliefs, etc., about diverse actualities such as services, goods, and topics. People intend to investigate the users' opinions on the online platform to achieve better performance. Regardless, the high-dimensional feature set in an online review study affects the interpretation of classification. Several studies have implemented different feature selection techniques; however, getting a high accuracy with a very minimal number of features is yet to be accomplished. This paper develops an effective hybrid approach based on an enhanced genetic algorithm (GA) and analysis of variance (ANOVA) to achieve this purpose. To beat the local minima convergence problem, this paper uses a unique two-phase crossover and impressive selection approach, gaining high exploration and fast convergence of the model. The use of ANOVA drastically reduces the feature size to minimize the computational burden of the model. Experiments are performed to estimate the algorithm performance using different conventional classifiers and algorithms like GA, Particle Swarm Optimization (PSO), Recursive Feature Elimination (RFE), Random Forest, ExtraTree, AdaBoost, GradientBoost, and XGBoost. The proposed novel approach gives impressive results using the Amazon Review dataset with an accuracy of 78.60 %, F1 score of 79.38 %, and an average precision of 0.87, and the Restaurant Customer Review dataset with an accuracy of 77.70 %, F1 score of 78.24 %, and average precision of 0.89 as compared to other existing algorithms. The result shows that the proposed model outperforms other algorithms with nearly 45 and 42% fewer features for the Amazon Review and Restaurant Customer Review datasets.

Many attempts have been made to overcome the challenges of automating textual emotion detection using different traditional deep learning models such as LSTM, GRU, and BiLSTM. But the problem with these models is that they need large datasets, massive computing resources, and a lot of time to train. Also, they are prone to forgetting and cannot perform well when applied to small datasets. In this paper, we aim to demonstrate the capability of transfer learning techniques to capture the better contextual meaning of the text and as a result better detection of the emotion represented in the text, even without a large amount of data and training time. To do this, we conduct an experiment utilizing a pre-trained model called EmotionalBERT, which is based on bidirectional encoder representations from transformers (BERT), and we compare its performance to RNN-based models on two benchmark datasets, with a focus on the amount of training data and how it affects the models' performance.

In the last decade, the need for a non-contact biometric model for recognizing candidates has increased, especially after the pandemic of COVID-19 appeared and spread worldwide. This paper presents a novel deep convolutional neural network (CNN) model that guarantees quick, safe, and precise human authentication via their poses and walking style. The concatenated fusion between the proposed CNN and a fully connected model has been formulated, utilized, and tested. The proposed CNN extracts the human features from two main sources: (1) human silhouette images according to model-free and (2) human joints, limbs, and static joint distances according to a model-based via a novel, fully connected deep-layer structure. The most commonly used dataset, CASIA gait families, has been utilized and tested. Numerous performance metrics have been evaluated to measure the system quality, including accuracy, specificity, sensitivity, false negative rate, and training time. Experimental results reveal that the proposed model can enhance recognition performance in a superior manner compared with the latest state-of-the-art studies. Moreover, the suggested system introduces a robust real-time authentication with any covariate conditions, scoring 99.8% and 99.6% accuracy in identifying casia (B) and casia (A) datasets, respectively.

Considering the huge volume of opinion texts published on various social networks, it is extremely difficult to peruse and use these texts. The automatic creation of summaries can be a significant help for the users of such texts. The current paper employs manifold learning to mitigate the challenges of the complexity and high dimensionality of opinion texts and the K-Means algorithm for clustering. Furthermore, summarization based on the concepts of the texts can improve the performance of the summarization system. The proposed method is unsupervised extractive, and summarization is performed based on the concepts of the texts using the multi-objective pruning approach. The main parameters utilized to perform multi-objective pruning include relevancy, redundancy, and coverage. The simulation results show that the proposed method outperformed the MOOTweetSumm method while providing an improvement of 11% in terms of the ROGUE-1 measure and an improvement of 9% in terms of the ROGUE-L measure.

In this paper, a novel optimization algorithm is proposed, called the Ladybug Beetle Optimization (LBO) algorithm, which is inspired by the behavior of ladybugs in nature when they search for a warm place in winter. The new proposed algorithm consists of three main parts: (1) determine the heat value in the position of each ladybug, (2) update the position of ladybugs, and (3) ignore the annihilated ladybug(s). The main innovations of LBO are related to both updating the position of the population, which is done in two separate ways, and ignoring the worst members, which leads to an increase in the search speed. Also, LBO algorithm is performed to optimize 78 well-known benchmark functions. The proposed algorithm has reached the optimal values of 73.3% of the benchmark functions and is the only algorithm that achieved the best solution of 20.5% of them. These results prove that LBO is substantially the best algorithm among other well-known optimization methods. In addition, two fundamentally different real-world optimization problems include the Economic-Environmental Dispatch Problem (EEDP) as an engineering problem and the Covid-19 pandemic modeling problem as an estimation and forecasting problem. The EEDP results illustrate that the proposed algorithm has obtained the best values in either the cost of production or the emission or even both, and the use of LBO for Covid-19 pandemic modeling problem leads to the least error compared to others.

Feature selection plays a very significant role for the success of pattern recognition and data mining. Based on the maximal relevance and minimal redundancy (mRMR) method, combined with feature subset, this paper proposes an improved maximal relevance and minimal redundancy (ImRMR) feature selection method based on feature subset. In ImRMR, the Pearson correlation coefficient and mutual information are first used to measure the relevance of a single feature to the sample category, and a factor is introduced to adjust the weights of the two measurement criteria. And an equal grouping method is exploited to generate candidate feature subsets according to the ranking features. Then, the relevance and redundancy of candidate feature subsets are calculated and the ordered sequence of these feature subsets is gained by incremental search method. Finally, the final optimal feature subset is obtained from these feature subsets by combining the sequence forward search method and the classification learning algorithm. Experiments are conducted on seven datasets. The results show that ImRMR can effectively remove irrelevant and redundant features, which can not only reduce the dimension of sample features and time of model training and prediction, but also improve the classification performance.

This paper presents a prototype filter design using the orthant optimization technique to assist a filter bank multicarrier (FBMC) modulation scheme of a NextG smart e-healthcare network framework. Low latency and very high reliability are one of the main requirements of a real-time e-healthcare system. In recent times, FBMC modulation has gotten more attention due to its spectral efficiency. The characteristics of a filter bank are determined by t's, prototype filter. A prototype filter cannot be designed to achieve an arbitrary time localization (for low latency) and frequency localization (spectral efficiency), as time and frequency spreading are conflicting goals. Hence, an optimum design needed to be achieved. In this paper, a constraint for perfect or nearly perfect reconstruction is formulated for prototype filter design and an orthant-based enriched sparse ℓ1-optimization method is applied to achieve the optimum performance in terms of higher availability of subcarrier spacing for the given requirement of signal-to-interference ratio. Larger subcarrier spacing ensures lower latency and better performance in real-time applications. The proposed FBMC system, based on an optimum design of the prototype filter, also supports a higher data rate as compared to traditional FBMC and OFDM systems, which is another requirement of real-time communication. In this paper, the solution for the different technical issues of physical layer design is provided. The presented modulation scheme through the proposed prototype filter-based FBMC can suppress the side lobe energy of the constituted filters up to large extent without compromising the recovery of the signal at the receiver end. The proposed system provides very high spectral efficiency; it can sacrifice large guard band frequencies to increase the subcarrier spacing to provide low-latency communication to support the real-time e-healthcare network.

Recently, interest in e-learning has increased rapidly owing to the lockdowns imposed by COVID-19. A major disadvantage of e-learning is the difficulty in maintaining concentration because of the limited interaction between teachers and students. The objective of this paper is to develop a methodology to predict e-learners' concentration by applying recurrent neural network models to eye gaze and facial landmark data extracted from e-learners' video data. One hundred eighty-four video data of ninety-two e-learners were obtained, and their frame data were extracted using the OpenFace 2.0 toolkit. Recurrent neural networks, long short-term memory, and gated recurrent units were utilized to predict the concentration of e-learners. A set of comparative experiments was conducted. As a result, gated recurrent units exhibited the best performance. The main contribution of this paper is to present a methodology to predict e-learners' concentration in a natural e-learning environment.

One of the most effective deterrent methods is using face masks to prevent the spread of the virus during the COVID-19 pandemic. Deep learning face mask detection networks have been implemented into COVID-19 monitoring systems to provide effective supervision for public areas. However, previous works have limitations: the challenge of real-time performance (i.e., fast inference and low accuracy) and training datasets. The current study aims to propose a comprehensive solution by creating a new face mask dataset and improving the YOLOv5 baseline to balance accuracy and detection time. Particularly, we improve YOLOv5 by adding coordinate attention (CA) module into the baseline backbone following two different schemes, namely YOLOv5s-CA and YOLOV5s-C3CA. In detail, we train three models with a Kaggle dataset of 853 images consisting of three categories: without a mask "NM," with mask "M," and incorrectly worn mask "IWM" classes. The experimental results show that our modified YOLOv5 with CA module achieves the highest accuracy mAP@0.5 of 93.9% compared with 87% of baseline and detection time per image of 8.0 ms (125 FPS). In addition, we build an integrated system of improved YOLOv5-CA and auto-labeling module to create a new face mask dataset of 7110 images with more than 3500 labels for three categories from YouTube videos. Our proposed YOLOv5-CA and the state-of-the-art detection models (i.e., YOLOX, YOLOv6, and YOLOv7) are trained on our 7110 images dataset. In our dataset, the YOLOv5-CA performance enhances with mAP@0.5 of 96.8%. The results indicate the enhancement of the improved YOLOv5-CA model compared with several state-of-the-art works.

This paper proposes a deep learning model that is robust and capable of handling highly uncertain inputs. The model is divided into three phases: creating a dataset, creating a neural network based on the dataset, and retraining the neural network to handle unpredictable inputs. The model utilizes entropy values and a non-dominant sorting algorithm to identify the candidate with the highest entropy value from the dataset. This is followed by merging the training set with adversarial samples, where a mini-batch of the merged dataset is used to update the dense network parameters. This method can improve the performance of machine learning models, categorization of radiographic images, risk of misdiagnosis in medical imaging, and accuracy of medical diagnoses. To evaluate the efficacy of the proposed model, two datasets, MNIST and COVID, were used with pixel values and without transfer learning. The results showed an increase of accuracy from 0.85 to 0.88 for MNIST and from 0.83 to 0.85 for COVID, which suggests that the model successfully classified images from both datasets without using transfer learning techniques.