Search results

Most previous studies predicting review helpfulness ignored the significance of deep features embedded in review text and instead relied on hand-crafted features. Hand-crafted and deep features have the advantages of high interpretability and predictive accuracy. This study aims to propose a novel review helpfulness prediction model that uses deep learning (DL) techniques to consider the complementarity between hand-crafted and deep features.

First, an advanced convolutional neural network was applied to extract deep features from unstructured review text. Second, this study used previous studies to extract hand-crafted features that impact the helpfulness of reviews and enhance their interpretability. Third, this study incorporated deep and hand-crafted features into a review helpfulness prediction model and evaluated its performance using the Yelp.com data set. To measure the performance of the proposed model, this study used 2,417,796 restaurant reviews.

Extensive experiments confirmed that the proposed methodology performs better than traditional machine learning methods. Moreover, this study confirms through an empirical analysis that combining hand-crafted and deep features demonstrates better prediction performance.

To the best of the authors’ knowledge, this is one of the first studies to apply DL techniques and use structured and unstructured data to predict review helpfulness in the restaurant context. In addition, an advanced feature-fusion method was adopted to better use the extracted feature information and identify the complementarity between features.

大多数先前预测评论有用性的研究忽视了嵌入在评论文本中的深层特征的重要性, 而主要依赖手工制作的特征。手工制作和深层特征具有高解释性和预测准确性的优势。本研究提出了一种新颖的评论有用性预测模型, 利用深度学习技术来考虑手工制作特征和深层特征之间的互补性。

首先, 采用先进的卷积神经网络从非结构化的评论文本中提取深层特征。其次, 本研究利用先前研究中提取的手工制作特征, 这些特征影响了评论的有用性并增强了其解释性。第三, 本研究将深层特征和手工制作特征结合到一个评论有用性预测模型中, 并使用Yelp.com数据集对其性能进行评估。为了衡量所提出模型的性能, 本研究使用了2,417,796条餐厅评论。

广泛的实验验证了所提出的方法优于传统的机器学习方法。此外, 通过实证分析, 本研究证实了结合手工制作和深层特征可以展现出更好的预测性能。

据我们所知, 这是首个在餐厅评论预测中应用深度学习技术, 并结合了结构化和非结构化数据来预测评论有用性的研究之一。此外, 本研究采用了先进的特征融合方法, 更好地利用了提取的特征信息, 并识别了特征之间的互补性。

To support achieving sustainable development goals (SDGs), we integrated science, technology, engineering and math (STEM) and extended reality technologies into an artificial intelligence (AI) learning activity. We developed Feature City to facilitate students' learning of AI concepts. This study aimed to explore students' learning outcomes and behaviors when using Feature City.

Junior high school students were the subjects who used Feature City in an AI learning activity. The learning activity consisted of 90-min sessions once per week for five weeks. Before the learning activity, the teacher clarified the learning objectives and administered a pretest. The teacher then instructed the students on the features, supervised learning and unsupervised learning units. After the learning activity, the teacher conducted a posttest. We analyzed the students' prior knowledge and learning performance by evaluating their pretest and posttest results and observing their learning behaviors in the AI learning activity.

(1) Students used Feature City to learn AI concepts to improve their learning outcomes. (2) Female students learned more effectively with Feature City than male students. (3) Male students were more likely than female students to complete the learning tasks in Feature City the first time they used it.

Within SDGs, this study used STEM and extended reality technologies to develop Feature City to engage students in learning about AI. The study examined how much Feature City improved students' learning outcomes and explored the differences in their learning outcomes and behaviors. The results showed that students' use of Feature City helped to improve their learning outcomes. Female students achieved better learning outcomes than their male counterparts. Male students initially exhibited a behavioral pattern of seeking clarification and error analysis when learning AI education, more so than their female counterparts. The findings can help teachers adjust AI education appropriately to match the tutorial content with students' AI learning needs.

Software defect prediction (SDP) is a critical aspect of software quality assurance, aiming to identify and manage potential defects in software systems. In this paper, we have proposed a novel hybrid approach that combines Gray Wolf Optimization with Feature Selection (GWOFS) and multilayer perceptron (MLP) for SDP. The GWOFS-MLP hybrid model is designed to optimize feature selection, ultimately enhancing the accuracy and efficiency of SDP. Gray Wolf Optimization, inspired by the social hierarchy and hunting behavior of gray wolves, is employed to select a subset of relevant features from an extensive pool of potential predictors. This study investigates the key challenges that traditional SDP approaches encounter and proposes promising solutions to overcome time complexity and the curse of the dimensionality reduction problem.

The integration of GWOFS and MLP results in a robust hybrid model that can adapt to diverse software datasets. This feature selection process harnesses the cooperative hunting behavior of wolves, allowing for the exploration of critical feature combinations. The selected features are then fed into an MLP, a powerful artificial neural network (ANN) known for its capability to learn intricate patterns within software metrics. MLP serves as the predictive engine, utilizing the curated feature set to model and classify software defects accurately.

The performance evaluation of the GWOFS-MLP hybrid model on a real-world software defect dataset demonstrates its effectiveness. The model achieves a remarkable training accuracy of 97.69% and a testing accuracy of 97.99%. Additionally, the receiver operating characteristic area under the curve (ROC-AUC) score of 0.89 highlights the model’s ability to discriminate between defective and defect-free software components.

Experimental implementations using machine learning-based techniques with feature reduction are conducted to validate the proposed solutions. The goal is to enhance SDP’s accuracy, relevance and efficiency, ultimately improving software quality assurance processes. The confusion matrix further illustrates the model’s performance, with only a small number of false positives and false negatives.

Hexahedral meshing is one of the most important steps in performing an accurate simulation using the finite element analysis (FEA). However, the current hexahedral meshing method in the industry is a nonautomatic and inefficient method, i.e. manually decomposing the model into suitable blocks and obtaining the hexahedral mesh from these blocks by mapping or sweeping algorithms. The purpose of this paper is to propose an almost automatic decomposition algorithm based on the 3D frame field and model features to replace the traditional time-consuming and laborious manual decomposition method.

The proposed algorithm is based on the 3D frame field and features, where features are used to construct feature-cutting surfaces and the 3D frame field is used to construct singular-cutting surfaces. The feature-cutting surfaces constructed from concave features first reduce the complexity of the model and decompose it into some coarse blocks. Then, an improved 3D frame field algorithm is performed on these coarse blocks to extract the singular structure and construct singular-cutting surfaces to further decompose the coarse blocks. In most modeling examples, the proposed algorithm uses both types of cutting surfaces to decompose models fully automatically. In a few examples with special requirements for hexahedral meshes, the algorithm requires manual input of some user-defined cutting surfaces and constructs different singular-cutting surfaces to ensure the effectiveness of the decomposition.

Benefiting from the feature decomposition and the 3D frame field algorithm, the output blocks of the proposed algorithm have no inner singular structure and are suitable for the mapping or sweeping algorithm. The introduction of internal constraints makes 3D frame field generation more robust in this paper, and it can automatically correct some invalid 3–5 singular structures. In a few examples with special requirements, the proposed algorithm successfully generates valid blocks even though the singular structure of the model is modified by user-defined cutting surfaces.

The proposed algorithm takes the advantage of feature decomposition and the 3D frame field to generate suitable blocks for a mapping or sweeping algorithm, which saves a lot of simulation time and requires less experience. The user-defined cutting surfaces enable the creation of special hexahedral meshes, which was difficult with previous algorithms. An improved 3D frame field generation method is proposed to correct some invalid singular structures and improve the robustness of the previous methods.

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.

Cracks are prevalent signs of pavement distress found on highways globally. The use of artificial intelligence (AI) and deep learning (DL) for crack detection is increasingly considered as an optimal solution. Consequently, this paper introduces a novel, fully connected, optimised convolutional neural network (CNN) model using feature selection algorithms for the purpose of detecting cracks in highway pavements.

To enhance the accuracy of the CNN model for crack detection, the authors employed a fully connected deep learning layers CNN model along with several optimisation techniques. Specifically, three optimisation algorithms, namely adaptive moment estimation (ADAM), stochastic gradient descent with momentum (SGDM), and RMSProp, were utilised to fine-tune the CNN model and enhance its overall performance. Subsequently, the authors implemented eight feature selection algorithms to further improve the accuracy of the optimised CNN model. These feature selection techniques were thoughtfully selected and systematically applied to identify the most relevant features contributing to crack detection in the given dataset. Finally, the authors subjected the proposed model to testing against seven pre-trained models.

The study's results show that the accuracy of the three optimisers (ADAM, SGDM, and RMSProp) with the five deep learning layers model is 97.4%, 98.2%, and 96.09%, respectively. Following this, eight feature selection algorithms were applied to the five deep learning layers to enhance accuracy, with particle swarm optimisation (PSO) achieving the highest F-score at 98.72. The model was then compared with other pre-trained models and exhibited the highest performance.

With an achieved precision of 98.19% and F-score of 98.72% using PSO, the developed model is highly accurate and effective in detecting and evaluating the condition of cracks in pavements. As a result, the model has the potential to significantly reduce the effort required for crack detection and evaluation.

The proposed method for enhancing CNN model accuracy in crack detection stands out for its unique combination of optimisation algorithms (ADAM, SGDM, and RMSProp) with systematic application of multiple feature selection techniques to identify relevant crack detection features and comparing results with existing pre-trained models.

Single-shot multi-category clothing recognition and retrieval play a crucial role in online searching and offline settlement scenarios. Existing clothing recognition methods based on RGBD clothing images often suffer from high-dimensional feature representations, leading to compromised performance and efficiency.

To address this issue, this paper proposes a novel method called Manifold Embedded Discriminative Feature Selection (MEDFS) to select global and local features, thereby reducing the dimensionality of the feature representation and improving performance. Specifically, by combining three global features and three local features, a low-dimensional embedding is constructed to capture the correlations between features and categories. The MEDFS method designs an optimization framework utilizing manifold mapping and sparse regularization to achieve feature selection. The optimization objective is solved using an alternating iterative strategy, ensuring convergence.

Empirical studies conducted on a publicly available RGBD clothing image dataset demonstrate that the proposed MEDFS method achieves highly competitive clothing classification performance while maintaining efficiency in clothing recognition and retrieval.

This paper introduces a novel approach for multi-category clothing recognition and retrieval, incorporating the selection of global and local features. The proposed method holds potential for practical applications in real-world clothing scenarios.

The purpose of this paper is to present an automatic approach for mesh sizing field generation of complicated  computer-aided design (CAD) models.

In this paper, the authors present an automatic approach for mesh sizing field generation. First, a source point extraction algorithm is applied to capture curvature and proximity features of CAD models. Second, according to the distribution of feature source points, an octree background mesh is constructed for storing element size value. Third, mesh size value on each node of background mesh is calculated by interpolating the local feature size of the nearby source points, and then, an initial mesh sizing field is obtained. Finally, a theoretically guaranteed smoothing algorithm is developed to restrict the gradient of the mesh sizing field.

To achieve high performance, the proposed approach has been implemented in multithreaded parallel using OpenMP. Numerical results demonstrate that the proposed approach is remarkably efficient to construct reasonable mesh sizing field for complicated CAD models and applicable for generating geometrically adaptive triangle/tetrahedral meshes. Moreover, since the mesh sizing field is defined on an octree background mesh, high-efficiency query of local size value could be achieved in the following mesh generation procedure.

How to determine a reasonable mesh size for complicated CAD models is often a bottleneck of mesh generation. For the complicated models with thousands or even ten thousands of geometric entities, it is time-consuming to construct an appropriate mesh sizing field for generating high-quality mesh. A parallel algorithm of mesh sizing field generation with low computational complexity is presented in this paper, and its usability and efficiency have been verified.

The purpose of this study is to establish a method for accurately extracting torch and seam features. This will improve the quality of narrow gap welding. An adaptive deflection correction system based on passive light vision sensors was designed using the Halcon software from MVtec Germany as a platform.

This paper proposes an adaptive correction system for welding guns and seams divided into image calibration and feature extraction. In the image calibration method, the field of view distortion because of the position of the camera is resolved using image calibration techniques. In the feature extraction method, clear features of the weld gun and weld seam are accurately extracted after processing using algorithms such as impact filtering, subpixel (XLD), Gaussian Laplacian and sense region for the weld gun and weld seam. The gun and weld seam centers are accurately fitted using least squares. After calculating the deviation values, the error values are monitored, and error correction is achieved by programmable logic controller (PLC) control. Finally, experimental verification and analysis of the tracking errors are carried out.

The results show that the system achieves great results in dealing with camera aberrations. Weld gun features can be effectively and accurately identified. The difference between a scratch and a weld is effectively distinguished. The system accurately detects the center features of the torch and weld and controls the correction error to within 0.3mm.

An adaptive correction system based on a passive light vision sensor is designed which corrects the field-of-view distortion caused by the camera’s position deviation. Differences in features between scratches and welds are distinguished, and image features are effectively extracted. The final system weld error is controlled to 0.3 mm.