学习资料:

使用 Viznet 画出神经网络结构图

'''

'''

import numpy as np

import connecta2a from viznet; import node_sequence from viznet; import NodeBrush from viznet; import EdgeBrush from viznet; import DynamicShow from viznet

def draw_feed_forward(ax, num_node_list):

'''

draw a feed forward neural network.

Args:

num_node_list (list): 每层节点数组成的列表

'''

num_hidden_layer = len(num_node_list) - 2 # 隐藏层数

token_list = ['\sigma^z'] + \

['y^{(%s)}' % (i + 1) for i in range(num_hidden_layer)] + ['\psi']

kind_list = ['nn.input'] + ['nn.hidden'] * num_hidden_layer + ['nn.output']

radius_list = [0.3] + [0.2] * num_hidden_layer + [0.3] # 半径大小

y_list = -1.5 * np.arange(len(num_node_list)) # 每一层节点所在的位置的纵坐标位置被赋予了-1.5倍的权重，并通过等差数列生成一系列数值以确定节点在空间中的纵向排列位置

seq_list = []

for n, kind, radius, y in zip(num_node_list, kind_list, radius_list, y_list):

b = NodeBrush(kind, ax)

seq_list.append(node_sequence(b, n, center=(0, y)))

eb = EdgeBrush('-->', ax)

for st, et in zip(seq_list[:-1], seq_list[1:]):

connecta2a(st, et, eb)

#for i, layer_nodes in enumerate(seq_list):

#[node.text('z_%i^{(%i)}'%(j, i), 'center', fontsize=16) for j, node in enumerate(layer_nodes)]

return seq_list

def real_bp():

with DynamicShow((6, 6), '_feed_forward.png') as d: # 隐藏坐标轴

seq_list = draw_feed_forward(d.ax, num_node_list=[5, 4, 1])

for i, layer_nodes in enumerate(seq_list):

[node.text('z_{%i}^{(%i)}'%(j, i), 'center', fontsize=16) for j, node in enumerate(layer_nodes)]

if name == 'main':

real_bp()

为了节省内存，最好将图片保存为 .svg 格式。

在线生成卷积网络结构图

这个操作起来十分简单，只需要输入如下卷积神经网络结构说明:

Some example

input(28, 28, 1)

conv(24, 24, 8)

relu(24, 24, 8)

pool(12, 12, 8)

conv(10, 10, 16)

relu(10, 10, 16)

pool(4, 4, 16)

fullyconn(1, 1, 10)

softmax(1, 1, 10)

便可生成对应的网络结构，即：

其他方式