针对大容量数据集，随机取部分而不是全部数据进行训练即可获得基本相同的网

For large-capacity data sets, you can get basically the same network by randomly taking part but not all of the data for training! Therefore, you can use the method of gradually increasing the samples for network training, and use the convergence of network weights as the iterative control condition to achieve Iterative incremental sampling training

For large data sets, train with random portions instead of all the data to get essentially the same network! Therefore, the network training can be carried out gradually by increasing the sample, and the network weight convergence situation can be used as the iterative control condition, so as to realize the iterative incremental sampling training

In view of the large capacity data set, we can get almost the same network by randomly taking part of the data instead of all the data for training. Therefore, we can use the method of gradually increasing samples to train the network, and take the convergence of network weights as the iterative control condition, so as to realize the Iterative Incremental sampling training<br>