山…隹

摘要： from tensorflow.keras.callbacks import EarlyStopping import tensorflow as tf import time import numpy as np import matplotlib.pyplot as plt import sys 阅读全文

摘要： from tensorflow.keras.callbacks import EarlyStopping import tensorflow as tf import time import numpy as np import matplotlib.pyplot as plt import sys 阅读全文

摘要： 流程图： self-attention： 矩阵运算： Q*K： V*（QK）： 整体过程： 阅读全文

摘要： from tensorflow import keras import tensorflow as tf from tensorflow.keras import layers import numpy as np import os import matplotlib.pyplot as plt 阅读全文

摘要： （1）没有output_padding: # 当padding="valid"时 new_rows = (rows - 1) * strides[0] + kernel_size[0] new_cols = (cols - 1) * strides[1] + kernel_size[1] # 当pa 阅读全文

摘要： 两个分布的对比（这里是与标准正态分布对比）： import tensorflow as tf from tensorflow import keras import numpy as np import matplotlib.pyplot as plt from PIL import Image i 阅读全文

摘要： import tensorflow as tf from tensorflow import keras import numpy as np import matplotlib.pyplot as plt from PIL import Image import os from tensorflo 阅读全文

摘要： (x,y),(x_test,y_test) = tf.keras.datasets.cifar10.load_data() x = x/255. x = tf.cast(x,tf.float32) x_test = x_test/255. x_test = tf.cast(x_test,tf.flo 阅读全文

摘要： # 图像增强 inputs = tf.image.random_flip_left_right(inputs) # 图像随机翻转 inputs = tf.image.random_brightness(inputs,max_delta=0.1) # 图像随机光照 inputs = tf.image. 阅读全文

摘要： self.conv1 = tf.keras.layers.Conv2D(filter_num, (3, 3), strides=stride, padding='same',kernel_initializer=tf.keras.initializers.he_normal(stddev=0.02) 阅读全文