使用初始化权重

mydl/ch4/dropout.py

+import numpy as np
+from numpy import ndarray
+
+from mydl.ch3.operation import Operation
+
+
+class Dropout(Operation):
+
+    def __init__(self,
+                 keep_prob: float = 0.8):
+        super().__init__()
+        self.keep_prob = keep_prob
+
+    def _output(self, inference: bool) -> ndarray:
+        if inference:
+            return self.input_ * self.keep_prob
+        else:
+            self.mask = np.random.binomial(1, self.keep_prob,
+                                           size=self.input_.shape)
+            return self.input_ * self.mask
+
+    def _input_grad(self, output_grad: ndarray) -> ndarray:
+        return output_grad * self.mask

mydl/my_test/np_test.py

+import matplotlib.pyplot as plt
+import numpy as np
+
+mu, sigma = 0, 0.1  # 均值和标准差
+s = np.random.normal(mu, sigma, 1000)
+print(s)
+print(abs(mu - np.mean(s)) < 0.01)
+print(abs(sigma - np.std(s, ddof=1)) < 0.01)
+count, bins, ignored = plt.hist(s, 40, color='b')
+print(count, bins, ignored)
+plt.plot(bins,
+         1 / (sigma * np.sqrt(2 * np.pi)) * np.exp(- (bins - mu) ** 2 / (2 * sigma ** 2)),
+         linewidth=2,
+         color='r')
+plt.show()