Create_AI_Framework_In5Classes(Day2) 的ForwardPropagation.py代码：
# -*- coding: utf-8 -*-#完成从Input Layer,经过若干层的Hidden Layers,最后得出Output Layer的值import mathclass ForwardPropagation:def applyForwardPropagation(nodes, weights, instance):for i in range(len(nodes)):if nodes[i].get_is_bias_unit() == True:nodes[i].set_value = 1# 把数据输入到Input Layer#例如说处理instance = [0,1,1]for j in range(len(instance) - 1): #训练的时候只需要featuresvalue_of_feature = instance[j] #获得该条数据中每个Feature具体的值for k in range(len(nodes)):if j + 1 == nodes[k].get_index(): #索引为0的节点为Bias，所以从索引为1的Node开始nodes[k].set_value(value_of_feature)#Hidden Layer的处理for j in range(len(nodes)):if nodes[j].get_is_bias_unit() == False and nodes[j].get_level() > 0 :target_neuron_input = 0 #接受上一个Layer中所有的和自己相关的Neurons和Weights的乘积之和target_neuron_output = 0 #经过Non-Linearity后的输出，我们这里使用Sigmoid#获得当前Neuron的IDtarget_index = nodes[j].get_index()for k in range(len(weights)):#获得和当前的Neuron关联的Weightif target_index == weights[k].get_to_index():#获得该Weight的Valueweight_value = weights[k].get_value()#获得该Weight的来源的Neuron的IDfrom_index = weights[k].get_from_index()#获得该来源Neuron的Valuefor m in range(len(nodes)):#获得该ID的Neuronif from_index == nodes[m].get_index():#获得该Neuron的具体的Valuevalue_from_neuron = nodes[m].get_value()#把Weight和相应的Value相乘，然后累加target_neuron_input = target_neuron_input + (weight_value * value_from_neuron)#一个Weight只连接一个上一个Layer的Neuron，所以不需要继续循环真个神经网络的其它Neurons'break#从和Break对其出发，一共按了4次后退键，因为接下来是要应用Sigmoid多当前的Neuron的所有的输入累计后的值进行操作target_neuron_output = 1 / (1 + math.exp(- target_neuron_input))#接下来把输入值和当然Neuron采用Sigmoid Activation计算后的值设置进当前的Neuronnodes[j].set_input_value(target_neuron_input)nodes[j].set_value(target_neuron_output)
(3) 测试结果。
在入口程序Neuron_Network_Entry.py中测试结果。输出结果的值是最后一个节点的值，
得出此次Forward Propagation的输出结果，从所有的节点中找到最后一个节点，打印出预测的值。
Neuron_Network_Entry.py运行结果如下：
+1 V1 V2Hidden layer creation: 1 N[1][1] N[1][2] N[1][3] N[1][4] N[1][5] N[1][6] N[1][7] N[1][8]Hidden layer creation: 2 N[2][1] N[2][2] N[2][3] N[2][4]Hidden layer creation: 3 N[3][1] N[3][2]Output layer: OutputThe weight from 1 at layers[0] to 4 at layers[1] : -0.01719299063527102The weight from 1 at layers[0] to 5 at layers[1] : -0.8524173292229386The weight from 1 at layers[0] to 6 at layers[1] : 0.22699060934105253The weight from 1 at layers[0] to 7 at layers[1] : -0.18342643007293868The weight from 1 at layers[0] to 8 at layers[1] : 0.535174965756674The weight from 1 at layers[0] to 9 at layers[1] : -0.14676978791733708The weight from 1 at layers[0] to 10 at layers[1] : 0.3575707340850214The weight from 1 at layers[0] to 11 at layers[1] : -0.42137618665671717The weight from 2 at layers[0] to 4 at layers[1] : -0.8009065486052088The weight from 2 at layers[0] to 5 at layers[1] : 0.058917238487063095The weight from 2 at layers[0] to 6 at layers[1] : -0.42346508944034544The weight from 2 at layers[0] to 7 at layers[1] : 0.8426870154158392The weight from 2 at layers[0] to 8 at layers[1] : 0.32010217521550643The weight from 2 at layers[0] to 9 at layers[1] : -0.18659699268657703The weight from 2 at layers[0] to 10 at layers[1] : -0.21967241566753914The weight from 2 at layers[0] to 11 at layers[1] : -0.24400451550197744The weight from 4 at layers[1] to 13 at layers[2] : 1.012406950277446The weight from 4 at layers[1] to 14 at layers[2] : -0.7119667051463217The weight from 4 at layers[1] to 15 at layers[2] : 0.6123794505814086The weight from 4 at layers[1] to 16 at layers[2] : 0.20933909060981204The weight from 5 at layers[1] to 13 at layers[2] : 0.8295825393038667The weight from 5 at layers[1] to 14 at layers[2] : -0.18589793075961192The weight from 5 at layers[1] to 15 at layers[2] : -0.4965519410696049The weight from 5 at layers[1] to 16 at layers[2] : 0.8986794993436826The weight from 6 at layers[1] to 13 at layers[2] : -0.5419190030559935The weight from 6 at layers[1] to 14 at layers[2] : -0.030482481689729557The weight from 6 at layers[1] to 15 at layers[2] : -0.16049573458078903The weight from 6 at layers[1] to 16 at layers[2] : -0.2974003908293369The weight from 7 at layers[1] to 13 at layers[2] : 0.2390039732386664The weight from 7 at layers[1] to 14 at layers[2] : 0.5368392670597157The weight from 7 at layers[1] to 15 at layers[2] : -0.38067640003252334The weight from 7 at layers[1] to 16 at layers[2] : 0.08850351612527696The weight from 8 at layers[1] to 13 at layers[2] : -0.45517979672262143The weight from 8 at layers[1] to 14 at layers[2] : -0.48321818662131666The weight from 8 at layers[1] to 15 at layers[2] : 0.5723650651069483The weight from 8 at layers[1] to 16 at layers[2] : 0.20266673558402148The weight from 9 at layers[1] to 13 at layers[2] : 0.4648137370852401The weight from 9 at layers[1] to 14 at layers[2] : -0.9281727938087562The weight from 9 at layers[1] to 15 at layers[2] : -0.3137211374881055The weight from 9 at layers[1] to 16 at layers[2] : -0.9786522293599609The weight from 10 at layers[1] to 13 at layers[2] : 0.7513533393983693The weight from 10 at layers[1] to 14 at layers[2] : 0.6710274413316895The weight from 10 at layers[1] to 15 at layers[2] : 0.9971668767938549The weight from 10 at layers[1] to 16 at layers[2] : 0.34557762622192767The weight from 11 at layers[1] to 13 at layers[2] : -0.8777163395171548The weight from 11 at layers[1] to 14 at layers[2] : -0.05999121610430269The weight from 11 at layers[1] to 15 at layers[2] : -0.6652778238381963The weight from 11 at layers[1] to 16 at layers[2] : 0.08832878707869818The weight from 13 at layers[2] to 18 at layers[3] : -0.8324236408834976The weight from 13 at layers[2] to 19 at layers[3] : -0.7635884380604687The weight from 14 at layers[2] to 18 at layers[3] : 0.8092754640273179The weight from 14 at layers[2] to 19 at layers[3] : -0.09275930457566717The weight from 15 at layers[2] to 18 at layers[3] : -0.6719236795958695The weight from 15 at layers[2] to 19 at layers[3] : -0.8016890418335867The weight from 16 at layers[2] to 18 at layers[3] : 1.0029644245873488The weight from 16 at layers[2] to 19 at layers[3] : -0.8780918045481161The weight from 18 at layers[3] to 20 at layers[4] : -0.8007053462211116The weight from 19 at layers[3] to 20 at layers[4] : 0.9594412274529027Prediction: 0.4477145403917822Prediction: 0.44384068472048943Prediction: 0.44955657205136573Prediction: 0.44559590058743986
instances中有4条数据，从输入数据的角度，第1条数据是[0,0],第2条数据是
[0,1],第3条数据是[1,0],第4条数据是[1,1]。结果中第一条数据的预测值是Prediction: 0.4477145403917822，但实际结果是0；第2条数据的预测值是Prediction: 0.44384068472048943，但实际结果是1；第3条数据的预测值是Prediction:0.44955657205136573，但实际结果是1；第4条数据的预测值是Prediction: 0.44559590058743986，但实际结果是0；我们只运行了1次，使用Sigmoid激活函数发现第1列和第2列跟第3列结果的关系，第1次运行的结果是不准确的。在下一章节中，我们将从结果中反推执行的过程，看每个节点关联的权重对误差造成的影响，根据我们的算法调整这个影响，随着循环过程的进行，误差会越来越小。TensorFlow的可视化图中，开始的误差是50%左右，我们这里的开始误差也很大，根据结果的计算，要么是0，要么是1，而我们的预测值如0.4477145403917822等，随着程序的运行，误差越来越小，而要完成这个过程，要根据结果反推以前的每一步，看权重值是否需要调整，调整权重之后，下一次再运行，神经元具体的值也会改变，因为权重发生了改变，不断的调整将越来越接近目标。
