Data expansion and classification method and system

The invention provides a data expansion and classification method and system. The method comprises the following steps: improving a DCGAN model through adding a random inactivation layer in the DCGAN model, and carrying out the expansion of data needing to be enhanced through the improved DCGAN model; and designing a multi-scale compression excitation network model, carrying out model parameter reduction and dimension reduction improvement, and carrying out data classification after expansion through the improved model. Based on the improved DCGAN model, the model is more stable, the quality of the generated picture is better, and a rich data set is provided for classification. Based on the improvement, the calculation amount is reduced, and the classification accuracy is improved. 本发明提供一种数据扩充及分类的方法、系统，所述方法包括如下步骤：通过在DCGAN模型中加入随机失活层对DCGAN模型进行改进，并通过改进的DCGAN模型对需要增强的数据进行扩充；设计多尺度压缩激励网络模型并进行模型参数减少以及降维的改进，并通过改进的模型进行扩充后数据的分类。基于改进DCGAN模型，从而使模型更稳定，生成图片质量更好，为分类提供丰富数据集。基于改进从而降低计算量，提高分类准确率。