Systems and Soft Computing最新文献

Due to the lack of accuracy and difficulty to meet the actual requirements of the poor students' financial assistance in the management of smart colleges and universities. Therefore, a precise funding model for poor students is constructed on the basis of improved reptile search algorithm and short-term memory neural network. The performance of the algorithm is evaluated by test function, rank sum test and combinatorial model validation. In test function 5, the algorithm is 2.90E+01±6.04E-03, which is lower than the comparison algorithm, and begins to converge after about 10 iterations, and the convergence speed is significantly higher than that of the comparison algorithm. In the rank sum test, the experimental results of the comparison algorithm on most test functions are less than 5 %. In the combined model verification, the fitness result of the maximum convergence times was 0.2203 %, and the classification accuracy reached 98.7 %, which was better than the comparison model. The precise funding model of poor students proposed in this study has important application value in the management of smart colleges and universities, which can effectively improve the accuracy of poor students' funding and meet the actual needs. At the same time, the high accuracy and fast convergence of the model provide a new idea and method for smart university management.

With the rapid development of Virtual Reality (VR) technology, its application in the field of sports training is also receiving increasing attention. This study applies the Distance Likelihood Based Probabilistic Model Averaging (DLPMA) algorithm to the VR motion intelligent capture system, aiming to provide an efficient and accurate motion data collection method to improve existing sports training methods. Introduced the design and implementation of a VR motion intelligent capture system based on DLPMA algorithm, and applied it to sports training. By conducting comparative experiments with traditional training methods, the advantages of the system in motion capture accuracy, real-time performance, and user experience are verified. The research results indicate that the system can accurately capture the movements of athletes and provide timely feedback to users, providing an effective auxiliary means for sports training. Although the system has shown good performance in sports training, there are still some limitations. Future research can further optimize algorithms, enhance system stability and flexibility, to meet a wider range of sports training needs.

With the increasing demand for electricity, predicting user electricity demand has become an essential task. The electricity demand characteristics of users in the electricity market are different, so it is necessary to classify and predict users. Aiming at the above problems, a classified forecasting model of electricity demand based on recent consumption, frequency, monetary (RFM), K-means, XGBoost and dynamic time warping (DTW) algorithm is proposed. The experiment showcases that among the electricity consumption of commercial users, the first type of load has the lowest proportion in autumn, at around 18.6 %; The second type of load has the highest proportion in autumn, about 81.3 %; Accurate classification has been made for the consuming quantity of electricity of commercial users. The average error in the forecasting results of the RFM-KM-XGboost model and the actual value of commercial electricity demand is about 0.07 kW; The average errors between the forecasting results of SVM model and RF model and the true values are about 0.2 kW and 0.14 kW, respectively; It indicates that the forecasting error of the RFM-KM-XGBoost model is smaller. The above results indicate that the RFM-KM-XGBoost model can extract users' electricity demand characteristics by classifying user types and load types, and make more accurate predictions of electricity demand for different types of users.

This paper presents a new study of predicting patients' breast cancer treatment protocol and the corresponding treatment cycle based on machine learning algorithms. The data used were collected at Mohammed VI Hospital in Morocco, and it contains patient information with two targets (protocol and treatment cycle).

After preparing the data and testing several machine learning algorithms, two models were developed: The first one, based on Gradient Boosting Classifier algorithm, successfully classified patient treatment protocols with an overall accuracy of 64 % across all categories and an impressive 94 % accuracy for the mode category, widely adopted in the hospital. The second model, based on Random Forest Regressor algorithm, which integrates the results of the first model during the training, predicted the treatment cycle of patients with a Root Mean Square Error (RMSE) score of 0.050 and a Mean Absolute Percentage Error (MAPE) score of 0.020. Furthermore, feature importance analysis was performed to highlight the importance of variables, and show the positive influence of some variables on the models.

Finally, this study can help doctors quickly make decisions about the treatment needed for each patient and also gives an idea of which molecules should exist in the hospital stock based on the patient's treatment cycle predicted.

To address the challenge of users selecting rich tourism resources, this study proposes a model for cultural tourism attraction recommendation using an optimized weighted association rule algorithm. This model includes time and season weight for tourist attraction recommendations. This model proposes improvement methods to address some inherent issues in traditional tourism recommendation models. Firstly, it constructed a recommendation model for cultural tourism attractions, and then optimized the weighted association rule algorithm by incorporating dynamic time weights. It takes into account the user's intended time in the recommendation outcome. Moreover, it incorporated seasonal weights to optimize the weighted rule algorithm for factors such as user travel time and the attractions' peak season during the recommendation process. The experiment indicates that the F1 value of the improved algorithm model proposed in this study reaches 0.952, the accuracy reaches 0.985, the area under the curve reaches 0.955, the Recall value reaches 0.812, and the fitting degree reaches 0.971. The results suggest that the proposed cultural tourism attraction recommendation model, based on an optimized weighted association algorithm, performs well in recommending tourist destinations. This model can have a positive impact on the development of the tourism industry.

This research presents a novel approach for plant disease identification utilizing Convolutional Neural Networks (CNNs) and the PYNQ FPGA platform. The study leverages the parallel processing capabilities of FPGAs to accelerate CNN inference, aiming to enhance the efficiency of plant disease detection in agricultural settings. The implementation involves optimizing the CNN architecture for deployment on the PYNQ FPGA, considering factors such as image size and learning rates. Through experimentation, the research refines hyper parameters, achieving improved accuracy and F1 scores. Visualizations using heat maps highlight the CNN's reliance on color, shape, and texture for feature extraction in disease identification. The integration of FPGA technology demonstrates promising advancements in real-time, high-performance plant disease classification, offering potential benefits for precision agriculture and crop management. This research contributes to the growing field of FPGA-accelerated deep learning applications in agro technology, addressing challenges in plant health monitoring and fostering sustainable agricultural practices.

This study aims to detect and count people using impulse radio ultra-wideband radar and machine learning algorithms. However, the data quality, difficulty distinguishing human signals from noise and clutter, and instances where human presence is not detected make it challenging to count multiple humans. To overcome these challenges, we apply wavelet transformation to reduce signal size and use simple moving averages to eliminate noise. Next, we create features based on statistical and entropic properties of the signal and apply several classification algorithms, including ANN, Random Forest, KNN, XGBOOST, and multiple linear regression, to predict the number of people present. Our findings reveal that using the ANN classifier with the Daubechies 4 (db4) wavelet provides better results than other classifiers, with an accuracy rate of 99%. Additionally, filtering the data improves accuracy, and labeling the data after extracting essential characteristics significantly improves the model’s accuracy.

The Fruit Fly Optimization Algorithm (FOA) has strong applicability, which can be optimized directly after the objective is determined not by building a complex model. Due to the problems of the algorithm such as easy prematureness, low solution accuracy, and easy to fall into local optimality. Therefore, the Gaussian Distribution Fruit Fly Optimization Algorithm (GaussFOA) based on Gaussian distribution was first proposed to solve the shortcomings of FOA. Then GaussFOA was applied to image segmentation processing. Finally, the experimental results were compared with FOA, the improved Fruit Fly Optimization Algorithm with Changing Step and Strategy (CSSFOA), and the Linear Generation Mechanism of Candidate Solution of Fruit Fly Optimization Algorithm (LGMSFOA). The results showed that GaussFOA had 100 % success rate compared with FOA, CSSFOA, and LGMSFOA under the same function. This algorithm also had the best finding mean and standard deviation. The low and high threshold division was compared in terms of the number of segmentation thresholds. The GaussFOA had the best value of both the average and the standard deviation of the search for merit. The segmentation results under high threshold were more obvious when compared with the segmentation results of low threshold GaussFOA. The image immunity of GaussFOA was 8.57 %, 10 %, and 29.97 % higher than that of FOA, Particle Swarm Optimization (PSO), and Genetic Algorithm (GA). This indicated that the model constructed based on GaussFOA had improved the image segmentation effect and stability compared with other algorithms. The findings of the research can offer a new path for the processing techniques of images.

In virtual reality interactive input systems, real-time and accurate spatial position and pose measurement is key to achieving a natural user interface. This study explores the application of inertial measurement units and binocular vision fusion technology in virtual reality interactive input systems, with the aim of improving the tracking accuracy of the system through optimized pose models and visual algorithms. A virtual reality measurement technology that integrates inertial measurement units and binocular vision is proposed by using an improved Kalman filtering algorithm to process inertial measurement units data, and combining it with the SURF algorithm optimized binocular vision system. Experimental results showed that fusion technology could reduce sensor noise and bias, improve the accuracy of pose estimation, and promote stable and robust motion tracking in virtual reality systems. In the angle range of -90 ° to 90 °, the average absolute error of the fusion system compared to the simple inertial measurement units pose calculation decreased from 3.371 ° to 1.369 ° In the experiment of measuring distance with a binocular vision system, the average absolute error value decreased to 1.532 mm, and the error range of the marker within the distance range of 500 mm to 650 mm was controlled within -1.3 mm to 1.2 mm. This study provides an effective solution for achieving high-precision virtual reality interactive input systems and is meaningful for the advancement of virtual reality technology.

With the advancement of education information, the scale of online education has been expanding, which brings challenges to students' learning path planning, i.e., course and learning method planning. To address the limitations of learning path planning such as insufficient personalization, the study proposes a learning path planning method based on learning path variability and ant colony optimization. First, dynamic time regularization is used to obtain learning path variability, and the K-means algorithm is used to classify students' learning types. Subsequently, an ant colony optimization algorithm is used to generate learning paths. Finally, the effectiveness of the method is tested. The results show that the loss value of the ant colony optimization algorithm converges to a minimum value of 0.1, which has the best stability of the loss function curve and the fastest convergence speed compared to other algorithms. Under the same experimental environment, the accuracy of the algorithm is as high as 0.9, which is conducive to the search for the optimal solution. The path planning method designed by the research can effectively grasp the learning characteristics and habits of students, and the accurate classification degree can reach 96.6%. With this learning path planning method, the average video learning time of students reaches a maximum of 80 min, while the average completion rate of students' course objectives is stable at 90%, which is about 20% higher than that of the GA-based learning path planning method. The method can significantly improve academic performance and educational outcomes. The method thus grasps the type of student learning, stimulates students' interest in learning, improves the effect of online learning, helps to promote education informatization and provides a boost to the deep reform of education.