2018.12.04 verification models and identification models

参考文章内容：Learning deep features from body and parts for person re-identification in camera networks

verification models and identification models:

The verification models [10–12] takepairs of images as the input and determine whether they belong to the same person or not, and the structure is shown in Fig. 1a. The verification models treat the person re-ID as a binary-class classification task and map two images of the same person into the nearby points; otherwise, the points are far apart. However, the verification models are not explicitly utilize re-ID labels.

To make full use of the re-ID labels, the identification models [13, 14], which treat the person re-ID as a multi-class recognition
task, are proposed, and the structure is shown in Fig. 1b. The identification models usually take holistic images as the input and determine whether images belong to their own labels. The identification models have shown great potentials on current large-scale person re-ID databases.Xiao et al. [15] jointly trained a identification model using multiple datasets and proposed a new dropout function to deal with a large number of classes.
 

 

假定你正在处理类属特征，并且没有查看分类变量在测试集中的分布。现在你想将 one hot encoding（OHE）应用到类属特征中。那么在训练集中将 OHE 应用到分类变量可能要面临的困难是什么？

 

A、分类变量所有的类别没有全部出现在测试集中

 

B、类别的频率分布在训练集和测试集是不同的

Skip gram 模型是在 Word2vec 算法中为词嵌入而设计的最优模型。以下哪一项描绘了 Skip gram 模型？

两个都是word2 ebd模型，模型A代表着CBOW,模型B 代表着skip-gram

对数损失函数不可能为负值

tanh relu sigmoid的函数的范围

i类错误 弃真   2类错误 取伪

给线性回归模型添加一个不重要的特征可能会造成R-quare的增加，特征空间增加特征之后，不论特征重不重要，R-quare通常会增加。

特征之间的相关系数不会因为特征加或者减一个数而改变

多元线性共线，高度正相关，高度负相关，线性回归时可以去除一个变量，或者用移除相关变量可能会导致信息损失。为了保留这些变量，我们可以使用带罚项的回归模型（如 ridge 或 lasso regression）。

t-SNE考虑最近邻点而进行数据的降维，所降的维可以在最近邻空间得到解释。