上一页 1 ··· 7 8 9 10 11 12 13 14 15 ··· 18 下一页
摘要: https://blog.csdn.net/qq_41936559/article/details/102699082 阅读全文
posted @ 2020-04-23 21:21 月夜_1 阅读(4674) 评论(0) 推荐(0) 编辑
摘要: 原创起个名字好难.JPG 最后发布于2018-09-18 22:20:34 阅读数 2118 收藏展开1.4 padding 在深度学习中,需要学会的一个基本卷积操作就是padding,padding是什么意思呢? 之前未使用padding时,假设我们的图片是6*6的,在使用3*3的filter之后 阅读全文
posted @ 2020-04-21 21:04 月夜_1 阅读(478) 评论(0) 推荐(0) 编辑
摘要: 数组新的shape属性应该要与原来的配套,如果等于-1的话,那么Numpy会根据剩下的维度计算出数组的另外一个shape属性值。 举个例子: x = np.array([[2, 0], [1, 1], [2, 3]]) 指定新数组行为3,列为,2,则: y = x.reshape(3,2) y Ou 阅读全文
posted @ 2020-04-20 09:54 月夜_1 阅读(805) 评论(0) 推荐(0) 编辑
摘要: 1.对一个一维向量 import numpy as npa = np.array([3, 1, 2, 4, 6, 1])b=np.argmax(a)#取出a中元素最大值所对应的索引,此时最大值位6,其对应的位置索引值为4,(索引值默认从0开始)print(b)#42.对2维向量(通常意义下的矩阵)a 阅读全文
posted @ 2020-04-20 09:51 月夜_1 阅读(929) 评论(0) 推荐(0) 编辑
摘要: keras.layers.convolutional.Conv1D(filters, kernel_size, strides=1, padding='valid', dilation_rate=1, activation=None, use_bias=True, kernel_initialize 阅读全文
posted @ 2020-04-20 09:22 月夜_1 阅读(2930) 评论(0) 推荐(0) 编辑
摘要: np_utils.to_categorical用于将标签转化为形如(nb_samples, nb_classes)的二值序列。假设num_classes = 10。如将[1,2,3,……4]转化成:[[0,1,0,0,0,0,0,0][0,0,1,0,0,0,0,0][0,0,0,1,0,0,0,0 阅读全文
posted @ 2020-04-15 13:24 月夜_1 阅读(3312) 评论(0) 推荐(0) 编辑
摘要: 官方文档:https://keras.io/layers/convolutional/#zeropadding2d https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.h 阅读全文
posted @ 2020-04-15 12:35 月夜_1 阅读(5754) 评论(0) 推荐(0) 编辑
摘要: 一、StratifiedKFold及KFold主要区别及函数参数KFold交叉采样:将训练/测试数据集划分n_splits个互斥子集,每次只用其中一个子集当做测试集,剩下的(n_splits-1)作为训练集,进行n_splits次实验并得到n_splits个结果。注:对于不能均等分的数据集,前n_s 阅读全文
posted @ 2020-04-15 11:44 月夜_1 阅读(1229) 评论(0) 推荐(0) 编辑
摘要: https://blog.csdn.net/qq_41635352/article/details/79963345 阅读全文
posted @ 2020-04-14 09:56 月夜_1 阅读(2320) 评论(0) 推荐(0) 编辑
摘要: 1. 关闭坐标刻度(plt 与 AxesSubplot) plt plt.xticks([]) plt.yticks([]) 关闭坐标轴: plt.axis('off') 注意,类似的这些操作若想起作用,需要将其置于 plt.show() 之前,plt.imshow() 之后。 对于 ax(matp 阅读全文
posted @ 2020-04-13 23:53 月夜_1 阅读(10214) 评论(0) 推荐(1) 编辑
上一页 1 ··· 7 8 9 10 11 12 13 14 15 ··· 18 下一页