New Advances in Deep Neural Networks for Big Data Classification and Prediction

In recent years, with the rapid development and popularization of Internet technology, the scale of data on the Internet has been rapidly increasing, leading to challenges such as high distribution and disorder in data organization. These challenges have reduced the utilization rate of Internet information resources. Addressing this issue, this study focuses on the scientific design and application of classification systems for Web big data to enable fast and accurate information retrieval. To tackle the challenges posed by high-dimensional and heterogeneous data, this research proposes advanced methods including Convolutional Neural Networks (CNN) and weighted sampling algorithms for big data classification and prediction. Experimental evaluations were conducted on power image and medical image datasets. The results reveal that the proposed methods achieved classification accuracies of 90% and 87% for power and medical images, respectively. Furthermore, the weighted sampling algorithm showed a significant improvement with a classification accuracy of 87%, compared to the original method's 82%. It also demonstrated superior performance in precision (85<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>), recall (84<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>), and F1-score (84.5%). These findings highlight the effectiveness of CNNs and weighted sampling in enhancing classification accuracy and efficiency for big data applications, providing valuable insights and references for future research in this domain.