CSE 250B: Assignment 7
Xinyi Wang A53099589
March 9, 2016
1 Singular values versus eigenvalues.
(a)
M vi = i ui
(b)
M T ui = i vi
(c)
M T M vi = M T i ui
= i i vi
= i2 vi
M M T ui = M i vi
= i i ui
= i2 ui
(d)
Since we have:
M M T = U V T V U T
= U 2 U T

CSE 250B: Assignment 6
Xinyi Wang A53099589
March 1, 2016
1 Experiments with clustering
(a)
Table 1 shows the list of k-means clusters. The result makes sense a lot, for example, animals in the
10th cluster are all mammal hunters which move fast. They sha

CSE 250B: Assignment 5
Xinyi Wang A53099589
February 18, 2016
1 Voting Perceptrons
(a)
The decision boundary is shown in Figure 1. As we can see from the result, the decision boundary is
not linear.
Figure 1: Decision boundary resulted from the voted perc

CSE 250B: Assignment 4
Xinyi Wang A53099589
February 11, 2016
1 Regression problem
(a)
n
X
L
(i)
=
2(y (i) x(i) )(xj )
j
i=1
n
X (i) (i)
2L
=
2xj xk
j k
i=1
Define v1 . . . vp Rp as follows:
vji =
(i)
2xj
And we have:
Hjk =
2L
= vj vk
j k
Thus, the Hessia

CSE 250B: Assignment 3
Xinyi Wang A53099589
February 3, 2016
1 Bivariate Gaussians
(a)
2
=
2
1
0.25
=
0.25 0.25
(b)
Since y is equal to x, corr(X, Y ) = 1. And we have:
1
1
=
=
1
1
2 Plot bivariate Gaussians
(a)
The plot is shown in Figure 1:
Figure 1

CSE 250B: Assignment 2
Xinyi Wang A53099589
January 23, 2016
1 Text classification with naive Bayes.
(c) First, use train.label to calculate j . For each class, calculate the number of documents that
belong to the class and divide by the total document nu

CSE 250B: Assignment 1
Xinyi Wang A53099589
January 15, 2016
1 Prototype Selection.
(a)
To select a subset of the training set, I traverse the original set and see whether the current image
satisfies the criteria I set up. If it satisfies the criteria, th