Abstract This paper proposes an optimized Ant Colony Algorithm (ACA) model to cluster and mine the physical exercise behavior characteristics of college students. Recognizing that students’ unstructured exercise habits often lead to inefficacy or injury, the... This paper proposes an optimized Ant Colony Algorithm (ACA) model to cluster and mine the physical exercise behavior characteristics of college students. Recognizing that students’ unstructured exercise habits often lead to inefficacy or injury, the study aims to provide more accurate behavioral analysis. The model first constructs a behavior representation by tracking the centroid movement of students’ bodies in video sequences to generate feature vectors. It then employs an ACA, inspired by ants’ foraging behavior, to cluster these vectors efficiently. The algorithm is optimized to address common ACA issues like slow convergence and stagnation by improving pheromone updating and path selection. Experimental results on standard and custom datasets show the model achieves higher F-measure values, greater stability, and faster convergence compared to other algorithms, effectively avoiding local optima for superior clustering accuracy. Read More Read Less

Abstract This paper proposes a rough set-based data mining algorithm to enhance data integrity verification within cloud-based accounting information systems. It addresses security risks inherent in cloud storage, such as data breaches and corruption, which traditional... This paper proposes a rough set-based data mining algorithm to enhance data integrity verification within cloud-based accounting information systems. It addresses security risks inherent in cloud storage, such as data breaches and corruption, which traditional verification methods handle inefficiently due to high computational and bandwidth costs. The authors introduce a novel parallel verification algorithm that can simultaneously check data for single or multiple users, significantly reducing communication time and computational overhead compared to single point testing. Experimental results demonstrate that this approach, leveraging cloud computing’s parallel processing capabilities, greatly improves computational efficiency and reduces update costs while ensuring data completeness. The system also allows users to select different audit levels, balancing verification speed, precision, and cost. The research aims to provide a more secure and efficient framework for financial data management in the era of accounting informationization. Read More Read Less

Abstract In order to better assist swimming athletes in correcting their abnormal postures, we have developed a pose recognition technology based on depth images. This technology uses threshold algorithms for data preprocessing and Kalman filters to... In order to better assist swimming athletes in correcting their abnormal postures, we have developed a pose recognition technology based on depth images. This technology uses threshold algorithms for data preprocessing and Kalman filters to filter noise. We also employ Gaussian distribution functions to capture dynamic changes and utilize the SURF algorithm to remove blurry parts. This technology can help us better assist swimming athletes in correcting their abnormal postures and improve their daily training. Using the Euclidean distance method, we can accurately estimate the distance between two adjacent reference points and employ feedback monitoring techniques to correct improper postures. Through simulations, we have found that this new deep level image skeleton tracking technology can effectively capture the dynamics of athletes and accurately detect their poses, demonstrating high accuracy and stability. Read More Read Less