Research on Data Classification Algorithms Based on Deep Learning and Its Performance Evaluation

Abstract

This study aims to investigate the performance of data classification algorithms based on deep learning and conduct a comprehensive evaluation and comparative analysis. Convolutional Neural Networks (CNN) and Long Short-Term Memory networks (LSTM) were
selected as representative deep learning models, while Support Vector Machines (SVM) and K-Nearest Neighbors (KNN) were introduced as
traditional classification algorithms for comparison. Experiments were conducted using the MNIST handwritten digit dataset and the CIFAR-
10 image dataset. Data preprocessing steps, including normalization, standardization, and augmentation, were applied to prepare the data for
training and testing. The results demonstrate that CNN achieved classification accuracies of 98.5% on MNIST and 86.7% on CIFAR-10, with
LSTM following closely. Traditional algorithms performed relatively poorly. Deep learning models exhibited significant advantages in terms
of precision, recall, and F1-score, while also achieving shorter training times and stronger generalization capabilities. The findings reveal that
deep learning algorithms have clear superiority in complex data classification tasks and are particularly suitable for high-dimensional, nonlinear data scenarios.