作业2------代码来源于B站up主ladykaka007 https://space.bilibili.com/49109393?spm_id_from=333.788.b_765f7570696e666f.2 手把手教大家实现吴恩达深度学习作业第二周
#导入数据 import h5py #训练原始数据 train_data = h5py.File('C:/Users/tf/Desktop/assignment2/datasets/train_catvnoncat.h5','r') test_data = h5py.File('C:/Users/tf/Desktop/assignment2/datasets/test_catvnoncat.h5','r') for key in train_data.keys(): print(key) -------------------------------- list_classes train_set_x train_set_y print(train_data['train_set_x'].shape) print(train_data['train_set_y'].shape) print(train_data['list_classes'].shape) ------------------------------------------ (209, 64, 64, 3) (209,) (2,) print(test_data['test_set_x'].shape) print(test_data['test_set_y'].shape) #取出数据集 测试集 train_data_org = train_data['train_set_x'][:] train_labels_org = train_data['train_set_y'][:] test_data_org = test_data['test_set_x'][:] test_labels_org = test_data['test_set_y'][:] print("训练集图片:"+str(train_data_org.shape)) print("训练集标签:"+str(train_labels_org.shape)) print("测试集图片:"+str(test_data_org.shape)) print("测试集标签:"+str(test_labels_org.shape)) ----------------------------------------- 训练集图片:(209, 64, 64, 3) 训练集标签:(209,) 测试集图片:(50, 64, 64, 3) 测试集标签:(50,) #查看图片 import matplotlib.pyplot as plt #matplotlib inline #可以在线显示图片，后面不用加plt.show（）显示图片 plt.imshow(train_data_org[176]) #数据纬度的处理 m_train = train_data_org.shape[0] m_test = test_data_org.shape[0] train_data_tran = train_data_org.reshape(m_train,-1).T test_data_tran = test_data_org.reshape(m_test,-1).T print(m_train) print(m_test) print(train_data_tran.shape) print(test_data_tran.shape) # train_labels_tran = train_labels_org.reshape(1,train_labels_org.shape[0]) # test_labels_tran = test_labels_org.reshape(1,test_labels_org.shape[0]) import numpy as np train_labels_tran = train_labels_org[np.newaxis,:]#和上面一行一个意思 test_labels_tran = test_labels_org[np.newaxis,:]#和上面一行一个意思 print(train_labels_tran.shape) print(test_labels_tran.shape) ------------------------------------------ 209 50 (12288, 209) (12288, 50) (1, 209) (1, 50) #标准化数据 train_data_sta = train_data_tran / 255 test_data_sta = test_data_tran / 255 print(train_data_sta.shape) print(test_data_sta.shape) #定义sigmoid函数 def sigmoid(z): a =1 / (1 + np.exp(-z)) return a #初始化参数 n_dim = train_data_sta.shape[0] print(n_dim) w = np.zeros((n_dim,1)) b = 0 #定义前向传播函数 def propagate(w,b,X,y): #1.前向传播函数 z = np.dot(w.T,X) + b A = sigmoid(z) #2.代价函数 m = X.shape[1] J = -1/m * np.sum(y * np.log(A)+(1-y) * np.log(1-A)) #3.梯度下降 dw = 1/m * np.dot(X,(A-y).T) db = 1/m * np.sum(A-y) grands = {'dw':dw,'db':db} # return dw,db,J #返回参数太多，而且w，b平常都是一起参考的，所以建议用字典，这个不是必须的 return grands,J #优化部分 def optimize(w,b,X,y,alpha,n_iters,print_cost): costs =[] for i in range(n_iters): grands,J = propagate(w,b,X,y) dw = grands['dw'] db = grands['db'] w = w - alpha * dw b = b - alpha * db if i % 100 == 0: costs.append(J) if print_cost: print('n_iters is',i,'cost is',J) grands = {'dw':dw,'db':db} params = {'w':w,'b':b} return grands,params,costs #预测部分 def predict(w,b,X_test): z = np.dot(w.T,X_test) + b A = sigmoid(z) m = X_test.shape[1] y_pred = np.zeros((1,m)) for i in range(m): if A[:,i] >0.5: y_pred[:,i] = 1 else: y_pred[:,i] = 0 return y_pred #模型的整合 def model(w,b,X_train,y_train,X_test,y_test,alpha,n_iters,print_cost): grands,params,costs = optimize(w,b,X_train,y_train,alpha,n_iters,print_cost) w = params['w'] b= params['b'] y_pred_train = predict(w,b,X_train) y_pred_test = predict(w,b,X_test) print("the train acc is",np.mean(y_pred_train == y_train)*100,'%') print("the test acc is",np.mean(y_pred_test == y_test)*100,'%') d = { 'w':w, 'b':b, 'costs':costs, 'y_pred_train':y_pred_train, 'y_pred_test':y_pred_test, 'alpha':alpha } return d d = model(w,b,train_data_sta,train_labels_tran,test_data_sta,test_labels_tran,alpha=0.005,n_iters =2000,print_cost =True) ----------------------------------------------------- n_iters is 0 cost is 0.6931471805599453 n_iters is 100 cost is 0.5845083636993086 n_iters is 200 cost is 0.46694904094655476 n_iters is 300 cost is 0.37600686694802077 n_iters is 400 cost is 0.3314632893282513 n_iters is 500 cost is 0.30327306747438293 n_iters is 600 cost is 0.2798795865826048 n_iters is 700 cost is 0.26004213692587574 n_iters is 800 cost is 0.24294068467796623 n_iters is 900 cost is 0.22800422256726066 n_iters is 1000 cost is 0.21481951378449635 n_iters is 1100 cost is 0.20307819060644985 n_iters is 1200 cost is 0.1925442771670686 n_iters is 1300 cost is 0.18303333796883503 n_iters is 1400 cost is 0.17439859438448876 n_iters is 1500 cost is 0.16652139705400335 n_iters is 1600 cost is 0.15930451829756614 n_iters is 1700 cost is 0.15266732471296504 n_iters is 1800 cost is 0.1465422350398234 n_iters is 1900 cost is 0.14087207570310162 the train acc is 99.04306220095694 % the test acc is 70.0 %训练集上面效果是99.04%,测试集上面是70%,说明有点过拟合了,太过贴合训练标签了,毕竟我们用的是逻辑回归,效果已经不错了
plt.plot(d['costs']) plt.xlabel('per hundred iters') plt.ylabel('costs') index = 45 print('y is',test_labels_tran[0,index]) print('y prediction is ',int(d['y_pred_test'][0,index]))结果：
y is 0 y prediction is 0 plt.imshow(test_data_org[index]) alphas = [0.01,0.001,0.0001] for i in alphas: print('alpha',i) d = model(w,b,train_data_sta,train_labels_tran,test_data_sta,test_labels_tran,alpha=i,n_iters =2000,print_cost =False) print("="*20) plt.plot(d['costs'],label = str(i)) plt.xlabel('per hundred iters') plt.ylabel('cost') plt.legend() ------------------------------------------- alpha 0.01 the train acc is 99.52153110047847 % the test acc is 70.0 % ==================== alpha 0.001 the train acc is 91.38755980861244 % the test acc is 68.0 % ==================== alpha 0.0001 the train acc is 71.29186602870813 % the test acc is 40.0 % ==================== # 下面是自己网上找的图片来经行验证 fname ='C:/Users/tf/Desktop/cat2.jpg' image = plt.imread(fname) plt.imshow(image) #检查一下图片维度 image.shape from skimage import transform image_tran = transform.resize(image,(64,64,3)).reshape(64*64*3,1) image_tran.shape ------------------------- (12288, 1) y = predict(d['w'],d['b'],image_tran) print(int(y))