Accuracy

class paddle.fluid.metrics.Accuracy(name=None)[源代码]

该接口用来计算多个mini-batch的平均准确率。Accuracy对象有两个状态value和weight。Accuracy的定义参照 https://en.wikipedia.org/wiki/Accuracy_and_precision

参数:
  • name (str,可选) – 具体用法请参见 Name ,一般无需设置,默认值为None。

返回:初始化后的 Accuracy 对象

返回类型:Accuracy

代码示例

import paddle.fluid as fluid
# 假设有batch_size = 128
batch_size=128
accuracy_manager = fluid.metrics.Accuracy()
# 假设第一个batch的准确率为0.9
batch1_acc = 0.9
accuracy_manager.update(value = batch1_acc, weight = batch_size)
print("expect accuracy: %.2f, get accuracy: %.2f" % (batch1_acc, accuracy_manager.eval()))
# 假设第二个batch的准确率为0.8
batch2_acc = 0.8
accuracy_manager.update(value = batch2_acc, weight = batch_size)
#batch1和batch2的联合准确率为(batch1_acc * batch_size + batch2_acc * batch_size) / batch_size / 2
print("expect accuracy: %.2f, get accuracy: %.2f" % ((batch1_acc * batch_size + batch2_acc * batch_size) / batch_size / 2, accuracy_manager.eval()))
#重置accuracy_manager
accuracy_manager.reset()
#假设第三个batch的准确率为0.8
batch3_acc = 0.8
accuracy_manager.update(value = batch3_acc, weight = batch_size)
print("expect accuracy: %.2f, get accuracy: %.2f" % (batch3_acc, accuracy_manager.eval()))
update(value, weight)

该函数使用输入的(value, weight)来累计更新Accuracy对象的对应状态,更新方式如下:

\[\begin{split}\\ \begin{array}{l}{\text { self. value }+=\text { value } * \text { weight }} \\ {\text { self. weight }+=\text { weight }}\end{array} \\\end{split}\]
参数:
  • value (float|numpy.array) – mini-batch的正确率
  • weight (int|float) – mini-batch的大小

返回:无

eval()

该函数计算并返回累计的mini-batches的平均准确率。

返回:累计的mini-batches的平均准确率

返回类型:float或numpy.array