向量的最大努力分类算法

找到 SSD (sum of squared differences) 对每个“类”使用测试向量,并使用具有最少SSD的测试向量。

下面是一些代码:我添加了一个 0 到测试向量的末尾,因为它只有9位数字,而类只有10位。

CLASSES = [1,0,0,0,0,0,0,0,0,0
           0,0,0,0,0,0,0,0,0,1
           0,1,1,1,1,1,1,1,1,0
           0,1,0,0,0,0,0,0,0,0];

TEST = [0.8,0,0,0,0.6,0,0.1,0,0,0];

% Find the difference between the TEST vector and each row in CLASSES
difference = bsxfun(@minus,CLASSES,TEST);
% Class differences
class_diff = sum(difference.^2,2);
% Store the row index of the vector with the minimum difference from TEST
[val CLASS_ID] = min(class_diff);
% Display
disp(CLASSES(CLASS_ID,:))

为了说明目的, difference 如下所示:

 0.2    0   0   0   -0.6    0   -0.1    0   0   0
-0.8    0   0   0   -0.6    0   -0.1    0   0   1
-0.8    1   1   1    0.4    1    0.9    1   1   0
-0.8    1   0   0   -0.6    0   -0.1    0   0   0

每节课到测试的距离是这样的, class_diff 以下内容:

 0.41
 2.01
 7.61
 2.01

显然,第一场比赛是最好的,因为它的差别最小。

这和 Jacob 只做了 四 不同的距离测量:

• Euclidean distance
• City-block distance
• Cosine distance
• Chebychev distance

%%
CLASSES = [1,0,0,0,0,0,0,0,0,0
           0,0,0,0,0,0,0,0,0,1
           0,1,1,1,1,1,1,1,1,0
           0,1,0,0,0,0,0,0,0,0];

TEST = [0.8,0,0,0,0.6,0,0.1,0,0,0];

%%
% sqrt( sum((x-y).^2) )
euclidean = sqrt( sum(bsxfun(@minus,CLASSES,TEST).^2, 2) );

% sum( |x-y| )
cityblock = sum(abs(bsxfun(@minus,CLASSES,TEST)), 2);

% 1 - dot(x,y)/(sqrt(dot(x,x))*sqrt(dot(y,y)))
cosine = 1 - ( CLASSES*TEST' ./ (norm(TEST)*sqrt(sum(CLASSES.^2,2))) );

% max( |x-y| )
chebychev = max( abs(bsxfun(@minus,CLASSES,TEST)), [], 2 );

dist = [euclidean cityblock cosine chebychev];

%%
[minDist classIdx] = min(dist);

选择你喜欢的那个:)

一个简单的欧几里得距离算法就足够了。距离该点最小的班级将是你的候选人。

http://en.wikipedia.org/wiki/Euclidean_distance