Explore the vast range of titles available.

This study aims to address the problems of incomplete temporal feature capture and low high-dimensional data visualization in mathematical education resource analysis within smart classrooms. It constructs a mathematical education resource visualization analysis model integrating the Bidirectional Long Short-Term Memory Network (BiLSTM) and Multi-Head Attention (MHA) mechanism. Taking the 16-week mathematics learning data of 1200 high school students from the smart classroom platform (including more than 150,000 behavior records, teaching resource attributes, and learning performance indicators) as the research object, this study cleans and normalizes the data, adopts Pearson correlation coefficients to screen key features, and constructs a multi-dimensional temporal matrix. Second, it uses a multi-layer BiLSTM to extract bidirectional temporal dependencies of learning behaviors, and combines MHA (number of heads = 4, single-head dimension = 32) to enhance the weights of key behaviors such as wrong-question review and delayed review. Finally, the study employs Principal Component Analysis (PCA) and t-distributed Stochastic Neighbor Embedding (t-SNE) to achieve dimensionality reduction and visualization of high-dimensional features. Experimental comparisons with models such as the single-layer Long Short-Term Memory Network (LSTM) and BiLSTM show that the Multi-BiLSTM + MHA model achieves optimal performance. This model outperforms other models in accuracy, F1-score, and resource utilization rate, with a fast-training convergence speed. In student group clustering, the high-activity group shows significantly higher knowledge mastery levels, with good group differentiation. This model can intuitively present differences in learning patterns and has strong generalization ability. This study provides technical support for personalized mathematics teaching and precise resource allocation in smart classrooms.

Prediction based on time series has a wide range of applications. Due to the complex nonlinear and random distribution of time series data, the performance of learning prediction models can be reduced by the modeling bias or overfitting. This paper proposes a novel planar flow-based variational auto-encoder prediction model (PFVAE), which uses the long- and short-term memory network (LSTM) as the auto-encoder and designs the variational auto-encoder (VAE) as a time series data predictor to overcome the noise effects. In addition, the internal structure of VAE is transformed using planar flow, which enables it to learn and fit the nonlinearity of time series data and improve the dynamic adaptability of the network. The prediction experiments verify that the proposed model is superior to other models regarding prediction accuracy and proves it is effective for predicting time series data.

The objective of seepage pressure monitoring of earth and rock dams is to predict seepage pressure in order to avoid potential risks. However, existing models for predicting seepage pressure in earth and rock dams do not account for the numerous nonlinearities between seepage pressure and the factors that influence it. These models lack the accuracy and generalizability required for effective risk management. In order to address this issue, this paper puts forth a methodology for the prediction of seepage pressure in earth and rock dams. This methodology is based on the use of Convolutional Neural Networks (CNN), Long Short-Term Memory Networks (LSTM), and an attention mechanism. The method initially normalizes each influence factor and divides the dataset. Subsequently, it employs a Convolutional Neural Network (CNN) to extract features from the data. Long Short-Term Memory (LSTM) networks are particularly adept at handling non-smooth time series data, enabling the capture of the deep information embedded within seepage pressure data. Furthermore, the introduction of attention mechanisms allows for the extraction of key information, ultimately enhancing the prediction accuracy and stability. The analysis of engineering examples demonstrates that, in comparison with the single CNN-LSTM, LSTM, Transformer, and BP models, the MAE, MAPE, and RMSE of the proposed method in this paper at two measurement points are the smallest among the four models. The results demonstrate that, in comparison to the other three prediction models, the method exhibits superior prediction accuracy and enhanced stability, is capable of discerning the local variation characteristics of seepage pressure data, exhibits enhanced robustness, and provides a novel approach for accurate prediction and analysis of seepage pressure in earth and rock dams.

Since 3D medical imaging data is a string of sequential images, there is a strong correlation between consecutive images. Deep convolutional networks perform well in extracting spatial features, but are less capable for processing sequence data compared to recurrent convolutional networks. Therefore, we propose a long short-term memory and attention based generative adversarial network (LSTMAGAN) to realize super-resolution reconstruction of 3D medical image. Firstly, we use generative adversarial networks as the base model for super-resolution image reconstruction. Secondly, a long and short-term memory network, which specializes in dealing with long-term dependencies in sequential data, was used to process continuous sequential data of 3D medical images based on its ability to remember and forget information efficiently. Next, an attention gate is used to suppress the background noise information and improve the clarity of image features. Finally, the method proposed in this paper is applied on the Luna16 and BraTs2021 datasets. The experimental results show that the proposed method improves the PSNR and SSIM evaluation indexes compared with other comparative methods, respectively. Therefore, it can prove the advancement and effectiveness of the proposed method.

Accurate prediction of vehicle maintenance demands is crucial for sustaining vehicle use, optimizing performance, and minimizing ownership costs. However, current methods only predict maintenance demand for specific vehicle components and lack the capability to offer a comprehensive prediction of all maintenance demands. Furthermore, predicting vehicle maintenance demand must incorporate the impacts of various essential maintenance projects on subsequent demands. To address these challenges, we propose an innovative method for predicting vehicle all maintenance demands based on collaborative multiview time series modeling. Leveraging the interdependencies among vehicle maintenance projects across various time periods, we employ a temporal dependency learning approach utilizing a multi-attention mechanism. To enhance the interaction between distinct time points and temporal dependencies, we developed a dependency-aware learning algorithm that effectively integrates and weighs the information and dependencies at each time step, thereby improving the model’s ability to capture the complex relationships among maintenance projects over time. To capture the significant impact of key maintenance projects on future demands, we propose a module that leverages both long short-term memory networks and the attention mechanism. Experimental results on actual vehicle maintenance records confirm that the proposed model outperforms existing methods, demonstrating its efficacy and applicability in predicting vehicle maintenance demand.

In the context of worsening global water scarcity and prominent water pollution issues, precise water quality assessment and prediction are crucial for water resource management. Addressing the shortcomings of traditional methods in multi-parameter collaborative analysis, class imbalance handling, and nonlinear fitting, this study focuses on Heilongjiang Province, China, to construct a multi-parameter watershed water quality level prediction system comprising “data preprocessing—feature dimensionality reduction—Model Optimization” multi-parameter watershed water quality level prediction system, combining Principal Component Analysis (PCA), C4.5 decision tree, Backpropagation (BP) neural network, Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM) networks for the study. The study collected 31,289 samples from 23 monitoring stations in the region between May and October 2023. After balancing the data using SMOTE oversampling and retaining information through PCA dimension reduction, the data were input into the four models. The results showed that PCA-C4.5 achieved an total accuracy of 87.26% but was weak in identifying small samples, while PCA-BP achieved the best total accuracy of 94.52%. PCA-CNN (93.27%) was good at capturing local features, and PCA-LSTM (93.42%) was good at capturing temporal patterns. This study confirms the feasibility of comparing and optimizing multiple algorithms (including C4.5 decision tree, BP neural network, CNN, and LSTM combined with PCA dimensionality reduction), providing a new method for watershed monitoring. Future research could incorporate attention mechanisms or multi-source data to enhance model performance.

Arctic sea ice plays a significant role in climate systems, and its prediction is important for coping with global warming. Artificial intelligence (AI) has gained recent attention in various disciplines with the increasing use of big data. In recent years, the use of AI-based sea ice prediction, along with conventional prediction models, has drawn attention. This study proposes a new deep learning (DL)-based Arctic sea ice prediction model with a new perceptual loss function to improve both statistical and visual accuracy. The proposed DL model learned spatiotemporal characteristics of Arctic sea ice for sequence-to-sequence predictions. The convolutional neural network-based perceptual loss function successfully captured unique sea ice patterns, and the widely used loss functions could not use various feature maps. Furthermore, the input variables that are essential to accurately predict Arctic sea ice using various combinations of input variables were identified. The proposed approaches produced statistical outcomes with better accuracy and qualitative agreements with the observed data.

Incineration is the main measure to deal with the increasing amount of municipal solid waste (MSW) in China. However, the effective management of fly ash, which is rich in a range of heavy metals, remains a challenge. In this study, the effective prediction of MSW generation in Shanghai is carried out using the LSTM (Long and Short Term Memory) prediction model with nine influencing factors of MSW generation as input indicators. Meanwhile, based on the prediction results and the general measurement formula of heavy metal content in MSW incineration fly ash, this study reasonably estimates the amount of MSW generated and the heavy metal content of incineration fly ash in Shanghai, and predicts the environmental impact of heavy metal in MSW incineration fly ash based on the potential environmental risk formula. The experimental results show that the MAPE (mean absolute percentage error) value of the LSTM prediction model in this study is 5.43%, which is better than the other four structural methods. The results showed that Shanghai MSW generation will reach 15.43 million tons by 2030, and incineration will produce 370,000 tons of fly ash. Cd, Hg and Cu will be the top three environmental risks of heavy metal pollution by MSWI (municipal solid waste incineration) in Shanghai, which need to be given priority to control.

Accurate runoff prediction is crucial for urban planning and environmental protection, land use management, and flood prevention and mitigation. Given that meteorological changes and human activities in upstream basins significantly affect downstream flow variations, we propose a novel runoff prediction method. This method combines long short-term memory networks with a graph convolutional neural network framework and integrates physical hydrological information to effectively capture spatial dependencies. A weighted directed graph is employed to construct a neighbor matrix, which learns the complex topology and spatial correlations among nodes. The proposed method was rigorously tested across multiple basins and benchmarked against state-of-the-art models. The results indicate a substantial enhancement in runoff prediction performance. Notably, the Nash–Sutcliffe model efficiency coefficient improved by up to 30.68%, the mean absolute error decreased by up to 44.41%, the mean relative error reduced by up to 55.05%, and the root mean square error decreased by up to 30.43%. These improvements, however, varied across different hydrological stations. The model serves as an effective large-scale runoff forecasting tool by integrating global hydrological feedback, thereby improving the accuracy of hydrological forecasts for both the main streams and tributaries within the basin.