clear_samples_class_mlpT_clear_samples_class_mlpClearSamplesClassMlpClearSamplesClassMlpclear_samples_class_mlp (算子)

名称

clear_samples_class_mlpT_clear_samples_class_mlpClearSamplesClassMlpClearSamplesClassMlpclear_samples_class_mlp — 清除多层感知机的训练数据。

签名

描述

clear_samples_class_mlpclear_samples_class_mlpClearSamplesClassMlpClearSamplesClassMlpClearSamplesClassMlpclear_samples_class_mlp clears all training samples that have been added to the multilayer perceptron (MLP) MLPHandleMLPHandleMLPHandleMLPHandleMLPHandlemlphandle with add_sample_class_mlpadd_sample_class_mlpAddSampleClassMlpAddSampleClassMlpAddSampleClassMlpadd_sample_class_mlp or read_samples_class_mlpread_samples_class_mlpReadSamplesClassMlpReadSamplesClassMlpReadSamplesClassMlpread_samples_class_mlp。clear_samples_class_mlpclear_samples_class_mlpClearSamplesClassMlpClearSamplesClassMlpClearSamplesClassMlpclear_samples_class_mlp should only be used if the MLP is trained in the same process that uses the MLP for evaluation with evaluate_class_mlpevaluate_class_mlpEvaluateClassMlpEvaluateClassMlpEvaluateClassMlpevaluate_class_mlp or for classification with classify_class_mlpclassify_class_mlpClassifyClassMlpClassifyClassMlpClassifyClassMlpclassify_class_mlp。In this case, the memory required for the training samples can be freed with clear_samples_class_mlpclear_samples_class_mlpClearSamplesClassMlpClearSamplesClassMlpClearSamplesClassMlpclear_samples_class_mlp, and hence memory can be saved. In the normal usage, in which the MLP is trained offline and written to a file with write_class_mlpwrite_class_mlpWriteClassMlpWriteClassMlpWriteClassMlpwrite_class_mlp, it is typically unnecessary to call clear_samples_class_mlpclear_samples_class_mlpClearSamplesClassMlpClearSamplesClassMlpClearSamplesClassMlpclear_samples_class_mlp because write_class_mlpwrite_class_mlpWriteClassMlpWriteClassMlpWriteClassMlpwrite_class_mlp does not save the training samples, and hence the online process, which reads the MLP with read_class_mlpread_class_mlpReadClassMlpReadClassMlpReadClassMlpread_class_mlp, requires no memory for the training samples.

执行信息

• 多线程类型：可重入（与非独占算子并行运行）。
• 多线程作用域：全局（可从任何线程调用）。
• 未采用并行化处理。

此算子修改后续输入参数的状态：

• MLPHandleMLPHandleMLPHandleMLPHandleMLPHandlemlphandle

在执行此算子时，若该参数值需在多个线程间使用，则必须对其访问进行同步。

参数

结果

如果参数有效，算子 clear_samples_class_mlpclear_samples_class_mlpClearSamplesClassMlpClearSamplesClassMlpClearSamplesClassMlpclear_samples_class_mlp 返回值 2 ( H_MSG_TRUE )。如有必要，则抛出异常。

可能的前趋

train_class_mlptrain_class_mlpTrainClassMlpTrainClassMlpTrainClassMlptrain_class_mlp, write_samples_class_mlpwrite_samples_class_mlpWriteSamplesClassMlpWriteSamplesClassMlpWriteSamplesClassMlpwrite_samples_class_mlp

另见

create_class_mlpcreate_class_mlpCreateClassMlpCreateClassMlpCreateClassMlpcreate_class_mlp, clear_class_mlpclear_class_mlpClearClassMlpClearClassMlpClearClassMlpclear_class_mlp, add_sample_class_mlpadd_sample_class_mlpAddSampleClassMlpAddSampleClassMlpAddSampleClassMlpadd_sample_class_mlp, read_samples_class_mlpread_samples_class_mlpReadSamplesClassMlpReadSamplesClassMlpReadSamplesClassMlpread_samples_class_mlp

模块

基础