Search results

The purpose of this paper is to estimate the causal effect of minimum wages (MWs) on the wages, employment and hours of migrant workers in China, and to show their inter-relatedness and how employers can offset some of the costs through subtle adjustments. This paper also illustrates the importance of disaggregating by region and sex.

Causal estimates are provided through difference-in-differences (DID) analysis, and robustness checks through propensity score matching. The analysis is based on micro data at the individual level from the household survey on migrant workers by the National Population and Family Planning Commission, combined with macro data regarding municipalities’ population, GDP and employment information based on the China Economic Information Network database.

MW increases for those paid by the month increased the earnings of both low-wage males and females. However, males tend not to experience an adverse employment effect because part of the cost increase is offset by employers increasing their monthly hours of work. Hours of work do not increase for females, so they experience an adverse employment effect. This highlights the importance of examining cost offsets such as increases in hours of work, as well as analyzing effects separately for males and females.

The reason behind why employers offset some of the cost increase for males paid by the month by increasing their hours of work, but this cost-offsetting adjustment does not occur for females is uncertain.

For workers paid by the month, employers can offset some of the cost increase by increasing their hours of work, leading to no reductions in employment. But this adjustment occurs only for males. Hours are not increased for females, but they experience reductions in employment. Clearly, MW increases have adverse effects either in the form of employment reductions (for females) or increases in hours of work for the same monthly pay (for males).

This paper provides causal estimates through DID analysis and robustness checks through Propensity Score Matching, and also indicates how employers can offset the cost of MW increases by increasing hours for those paid by the month, resulting in no adverse employment effect for such workers, but an adverse employment effect when such an adjustment does not occur.

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.

For studies in educational data mining or learning Analytics, the prediction of student’s performance or early warning is one of the most popular research topics. However, research gaps indicate a paucity of research using machine learning and deep learning (DL) models in predictive analytics that include both behaviors and text analysis.

This study combined behavioral data and discussion board content to construct early warning models with machine learning and DL algorithms. In total, 680 course sections, 12,869 students and 14,951,368 logs were collected from a K-12 virtual school in the USA. Three rounds of experiments were conducted to demonstrate the effectiveness of the proposed approach.

The DL model performed better than machine learning models and was able to capture 51% of at-risk students in the eighth week with 86.8% overall accuracy. The combination of behavioral and textual data further improved the model’s performance in both recall and accuracy rates. The total word count is a more general indicator than the textual content feature. Successful students showed more words in analytic, and at-risk students showed more words in authentic when text was imported into a linguistic function word analysis tool. The balanced threshold was 0.315, which can capture up to 59% of at-risk students.

The results of this exploratory study indicate that the use of student behaviors and text in a DL approach may improve the predictive power of identifying at-risk learners early enough in the learning process to allow for interventions that can change the course of their trajectory.

Online learning is well-known by its flexibility of learning anytime and anywhere. However, how behavioral patterns tied to learning anytime and anywhere influence learning outcomes are still unknown.

This study proposed concepts of time and location entropy to depict students’ spatial-temporal patterns. A total of 5,221 students with 1,797,677 logs, including 485 on-the-job students and 4,736 full-time students, were analyzed to depict their spatial-temporal learning patterns, including the relationships between identified patterns and students’ learning performance.

Analysis results indicate on-the-job students took more advantage of anytime, anywhere than full-time students. Students with a higher tendency for learning anytime and a lower level of learning anywhere were more likely to have better outcomes. Gender did not show consistent findings on students’ spatial-temporal patterns, but partial findings could be supported by evidence in neural science or by cultural and geographical differences.

A more accurate approach for categorizing position and location might be considered. Some findings need more studies for further validation. Finally, future research can consider connections between other well-known performance predictors (such as financial situation, motivation, personality and major) and the type of learning patterns.

The findings gained from this study can help improve the understandings of students’ learning behavioral patterns and design as well as implement better online education programs.

This study proposed concepts of time and location entropy to identify successful spatial-temporal patterns of on-the-job and full-time students.

Critical thinking is considered important in psychological science because it enables students to make effective decisions and optimizes their performance. Aiming at the challenges and issues of understanding the student's critical thinking, the objective of this study is to analyze online discussion data through an advanced multi-feature fusion modeling (MFFM) approach for automatically and accurately understanding the student's critical thinking levels.

An advanced MFFM approach is proposed in this study. Specifically, with considering the time-series characteristic and the high correlations between adjacent words in discussion contents, the long short-term memory–convolutional neural network (LSTM-CNN) architecture is proposed to extract deep semantic features, and then these semantic features are combined with linguistic and psychological knowledge generated by the LIWC2015 tool as the inputs of full-connected layers to automatically and accurately predict students' critical thinking levels that are hidden in online discussion data.

A series of experiments with 94 students' 7,691 posts were conducted to verify the effectiveness of the proposed approach. The experimental results show that the proposed MFFM approach that combines two types of textual features outperforms baseline methods, and the semantic-based padding can further improve the prediction performance of MFFM. It can achieve 0.8205 overall accuracy and 0.6172 F1 score for the “high” category on the validation dataset. Furthermore, it is found that the semantic features extracted by LSTM-CNN are more powerful for identifying self-introduction or off-topic discussions, while the linguistic, as well as psychological features, can better distinguish the discussion posts with the highest critical thinking level.

With the support of the proposed MFFM approach, online teachers can conveniently and effectively understand the interaction quality of online discussions, which can support instructional decision-making to better promote the student's knowledge construction process and improve learning performance.

Educational data mining (EDM) and learning analytics, which are highly related subjects but have different definitions and focuses, have enabled instructors to obtain a holistic view of student progress and trigger corresponding decision-making. Furthermore, the automation part of EDM is closer to the concept of artificial intelligence. Due to the wide applications of artificial intelligence in assorted fields, the authors are curious about the state-of-art of related applications in Education.

This study focused on systematically reviewing 1,219 EDM studies that were searched from five digital databases based on a strict search procedure. Although 33 reviews were attempted to synthesize research literature, several research gaps were identified. A comprehensive and systematic review report is needed to show us: what research trends can be revealed and what major research topics and open issues are existed in EDM research.

Results show that the EDM research has moved toward the early majority stage; EDM publications are mainly contributed by “actual analysis” category; machine learning or even deep learning algorithms have been widely adopted, but collecting actual larger data sets for EDM research is rare, especially in K-12. Four major research topics, including prediction of performance, decision support for teachers and learners, detection of behaviors and learner modeling and comparison or optimization of algorithms, have been identified. Some open issues and future research directions in EDM field are also put forward.

Limitations for this search method include the likelihood of missing EDM research that was not captured through these portals.

This systematic review has not only reported the research trends of EDM but also discussed open issues to direct future research. Finally, it is concluded that the state-of-art of EDM research is far from the ideal of artificial intelligence and the automatic support part for teaching and learning in EDM may need improvement in the future work.

The main purpose of this paper is to develop a method to recognize the initial welding position for large-diameter pipeline automatically, and introduce the image processing based on pulse-coupled neural network (PCNN) which is adopted by the proposed method.

In this paper, a passive vision sensor is designed to capture weld seam images in real time. The proposed method contains two steps. The first step is to detect the rough position of the weld seam, and the second step is to recognize one of the solder joints from the local image and extract its centroid, which is regarded as the initial welding position. In each step, image segmentation and removal of small false regions based on PCNN are adopted to obtain the object regions; then, the traditional image processing theory is used for the subsequent processing.

The experimental results show the feasibility and real time of the proposed method. Based on vision sensing technology and PCNN, it is able to achieve the autonomous recognition of initial welding position in large-diameter pipeline welding.

The proposed method can greatly shorten the time of positioning the initial welding position and satisfy the automatic welding for large-diameter pipeline.

In the proposed method, the image pre-processing is based on PCNN, which is more robust and flexible in the complex welding environment. After that, traditional image processing theory is adopted for the subsequent processing, of which the processing speed is faster.

Fire smoke detection in petrochemical plant can prevent fire and ensure production safety and life safety. The purpose of this paper is to solve the problem of missed detection and false detection in flame smoke detection under complex factory background.

This paper presents a flame smoke detection algorithm based on YOLOv5. The target regression loss function (CIoU) is used to improve the missed detection and false detection in target detection and improve the model detection performance. The improved activation function avoids gradient disappearance to maintain high real-time performance of the algorithm. Data enhancement technology is used to enhance the ability of the network to extract features and improve the accuracy of the model for small target detection.

Based on the actual situation of flame smoke, the loss function and activation function of YOLOv5 model are improved. Based on the improved YOLOv5 model, a flame smoke detection algorithm with generalization performance is established. The improved model is compared with SSD and YOLOv4-tiny. The accuracy of the improved YOLOv5 model can reach 99.5%, which achieves a more accurate detection effect on flame smoke. The improved network model is superior to the existing methods in running time and accuracy.

Aiming at the actual particularity of flame smoke detection, an improved flame smoke detection network model based on YOLOv5 is established. The purpose of optimizing the model is achieved by improving the loss function, and the activation function with stronger nonlinear ability is combined to avoid over-fitting of the network. This method is helpful to improve the problems of missed detection and false detection in flame smoke detection and can be further extended to pedestrian target detection and vehicle running recognition.