python-3.x - 当输出标签是3D数组时如何使用SVM？

Question

问题很简单，我有一个 3D 图像，我想使用 SVM 对它们进行分割。所以我将输入和输出图像转换为3D numpy数组，现在我想使用SVM。但似乎clf.fit()不支持多维标签。那么如何训练标签是多维数组的模型呢？

一个简单的例子:

from sklearn import svm
x=[[0,0],[1,1]]
y=[[0,0],[1,1]]
clf=svm.SVC(gamma='scale')
clf.fit(x,y)

错误是:

Traceback (most recent call last):
  File "basic.py", line 5, in <module>
    clf.fit(x,y)
  File "/usr/local/lib/python3.5/dist-packages/sklearn/svm/base.py", line 149, in fit
    accept_large_sparse=False)
  File "/usr/local/lib/python3.5/dist-packages/sklearn/utils/validation.py", line 761, in check_X_y
    y = column_or_1d(y, warn=True)
  File "/usr/local/lib/python3.5/dist-packages/sklearn/utils/validation.py", line 797, in column_or_1d
    raise ValueError("bad input shape {0}".format(shape))
ValueError: bad input shape (2, 2)

Answer 1

您正在添加不同的 y 类标签，这就是它不起作用的原因。请参阅下面带有内嵌注释的解决方案。

from sklearn import svm
x=[[0,0],[1,1],[7,8]]
y=[0,1, 2]                 # class labels
clf=svm.SVC()              # clf=svm.SVC(gamma='scale') > gamma is auto. no need to add this.

print (clf.fit(x,y))

q = clf.predict([[2., 2.]])  # simple example to test prediction.

print ('array : %s ' % q)


# use of multiple class labes for y

x=[[0,0],[1,1]]
y=[[0,1],[0,2]]          # the value 2 is to show the difference in printed output.

# add here your `for item in x:` if both arrays are 3D. `for item in y:` needs
# indentation if you do.

for item in y:           # iters through the labeling list.
    print (item)
    clf=svm.SVC()

    print (clf.fit(x,item))

    q = clf.predict([[2., 2.]])

    print ('array : %s ' % q)

打印结果:

SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0,
  decision_function_shape=None, degree=3, gamma='auto', kernel='rbf',
  max_iter=-1, probability=False, random_state=None, shrinking=True,
  tol=0.001, verbose=False)
array : [1] 
[0, 1]
SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0,
  decision_function_shape=None, degree=3, gamma='auto', kernel='rbf',
  max_iter=-1, probability=False, random_state=None, shrinking=True,
  tol=0.001, verbose=False)
array : [1] 
[0, 2]
SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0,
  decision_function_shape=None, degree=3, gamma='auto', kernel='rbf',
  max_iter=-1, probability=False, random_state=None, shrinking=True,
  tol=0.001, verbose=False)
array : [2]