def Get_svm_model(parameter, X, y):
svm = SVM()
loss_histroy = svm.Train(
X, y, parameter[1], 1, parameter[0], 200, 1500, True)
VisualizeLoss(loss_histroy)
input("Enter any key to predict...")
return svm
def AutoGetPara(X_train, y_train, X_val, y_val):
learning_rate = [1e-8, 1e-7] # 学习率
regularization_strengths = [5e4, 1e5] # 正则化强度
best_parameter = None
best_val = -1 # 最优评分值
delta = 1 # SVM损失函数中的delta参数
batch_num = 200 # 小批量训练
num_iter = 1500 # 迭代次数
for i in learning_rate:
for j in regularization_strengths:
svm = SVM()
svm.Train(X_train, y_train, j, delta, i,
batch_num, num_iter, True) # 得到了权重W
y_pred = svm.Predict(X_val) # 利用权重W得到评分值
acc_val = np.mean(y_val == y_pred)
if best_val < acc_val:
best_val = acc_val
best_parameter = (i, j)
print('OK! Have been identified parameter! Best validation accuracy achieved during cross-validation: %f' % best_val)
return best_parameter
