CS229 Problem Set #2 1

CS 229, Summer 2019

Problem Set #2 Solutions
Yusu Qian (006239176)

Due Monday, July 29 at 11:59 pm on Gradescope.

Notes: (1) These questions require thought, but do not require long answers. Please be as
concise as possible. (2) If you have a question about this homework, we encourage you to post
your question on our Piazza forum, at http://piazza.com/stanford/summer2019/cs229. (3)
If you missed the first lecture or are unfamiliar with the collaboration or honor code policy,
please read the policy on the course website before starting work. (4) For the coding problems,
you may not use any libraries except those defined in the provided environment.yml file. In
particular, ML-specific libraries such as scikit-learn are not permitted. (5) To account for late
days, the due date is Monday, July 29 at 11:59 pm. If you submit after Monday, July 29 at
11:59 pm, you will begin consuming your late days. If you wish to submit on time, submit before
Monday, July 29 at 11:59 pm.
All students must submit an electronic PDF version of the written questions. We highly rec-
ommend typesetting your solutions via LATEX. All students must also submit a zip file of their
source code to Gradescope, which should be created using the make zip.py script. You should
make sure to (1) restrict yourself to only using libraries included in the environment.yml file,
and (2) make sure your code runs without errors. Your submission may be evaluated by the
auto-grader using a private test set, or used for verifying the outputs reported in the writeup.
CS229 Problem Set #2 2

Figure 1: 1b plot of data set 1 (left) and data set 2 (right)

1. [15 points] Logistic Regression: Training stability

In this problem, we will be delving deeper into the workings of logistic regression. The goal of
this problem is to help you develop your skills debugging machine learning algorithms (which
can be very different from debugging software in general).
We have provided an implementation of logistic regression in src/stability/stability.py,
and two labeled datasets A and B in src/stability/ds1 a.csv and src/stability/ds1 b.csv.
Please do not modify the code for the logistic regression training algorithm for this problem.
First, run the given logistic regression code to train two different models on A and B. You can
run the code by simply executing python stability.py in the src/stability directory.

(a) [2 points] What is the most notable difference in training the logistic regression model on
datasets A and B?
Answer:
The training algorithm converges on data set A quickly while can’t converge on data set B (at
least in a reasonable time).
(b) [5 points] Investigate why the training procedure behaves unexpectedly on dataset B, but
not on A. Provide hard evidence (in the form of math, code, plots, etc.) to corroborate
your hypothesis for the misbehavior. Remember, you should address why your explanation
does not apply to A.
Hint: The issue is not a numerical rounding or over/underflow error.
Answer:
We can see from the plots (on top of this page) that in data set 2, the two classes are perfectly
seperated. Thus, it makes θ increase continuously to increase the likelihood. In data set 1, the
data points are mixed to some extent.
(c) [5 points] For each of these possible modifications, state whether or not it would lead to
the provided training algorithm converging on datasets such as B. Justify your answers.
i. Using a different constant learning rate.
ii. Decreasing the learning rate over time (e.g. scaling the initial learning rate by 1/t2 ,
where t is the number of gradient descent iterations thus far).
iii. Linear scaling of the input features.
iv. Adding a regularization term kθk22 to the loss function.
CS229 Problem Set #2 3

v. Adding zero-mean Gaussian noise to the training data or labels.

Answer:
i. Changing the learning rate won’t help as it only changes the speed θ reaches the maximum
value allowed. The algorithm won’t converge on data set 2.

ii. The algorithm won’t converge on data set 2. The reason is similar to question i, as learning
rate is not what caused the inability to converge.

iii. The algorithm won’t converge on data set 2. Again, this problem is caused by the prefect
seperateness, not by input features alone.

iv. The algorithm will converge on data set 2. This method restricts θ from endless increasing.

v. The algorithm will converge on data set 2. This will avoid the situation where data points
are perfectly seperated.

(d) [3 points] Are support vector machines, vulnerable to datasets like B? Why or why not?
Give an informal justification.
Answer:
No, because they look for geometric margins ( and maximize them) instead of functional margins.
CS229 Problem Set #2 4

2. [22 points] Spam classification

In this problem, we will use the naive Bayes algorithm and an SVM to build a spam classifier.
In recent years, spam on electronic media has been a growing concern. Here, we’ll build a
classifier to distinguish between real messages, and spam messages. For this class, we will be
building a classifier to detect SMS spam messages. We will be using an SMS spam dataset
developed by Tiago A. Almedia and Jos Mara Gmez Hidalgo which is publicly available on
http://www.dt.fee.unicamp.br/~tiago/smsspamcollection 1
We have split this dataset into training and testing sets and have included them in this assignment
as src/spam/spam train.tsv and src/spam/spam test.tsv. See src/spam/spam readme.txt
for more details about this dataset. Please refrain from redistributing these dataset files. The
goal of this assignment is to build a classifier from scratch that can tell the difference the spam
and non-spam messages using the text of the SMS message.

(a) [5 points] Implement code for processing the the spam messages into numpy arrays that can
be fed into machine learning models. Do this by completing the get words, create dictionary,
and transform text functions within our provided src/spam.py. Do note the correspond-
ing comments for each function for instructions on what specific processing is required.
The provided code will then run your functions and save the resulting dictionary into
spam dictionary and a sample of the resulting training matrix into
spam sample train matrix.
In your writeup, report the vocabular size after the pre-processing step. You do not need
to include any other output for this subquestion.
Answer:
The size is 1758 when only lowercase is performed. I also removed punctuations and numbers,
after which the size is 1539.
(b) [10 points] In this question you are going to implement a naive Bayes classifier for spam
classification with multinomial event model and Laplace smoothing (refer to class notes
on Naive Bayes for details on Laplace smoothing in Section 2.3 of notes2.pdf).
Code your implementation by completing the fit naive bayes model and
predict from naive bayes model functions in src/spam/spam.py.
Now src/spam/spam.py should be able to train a Naive Bayes model, compute your predic-
tion accuracy and then save your resulting predictions to spam naive bayes predictions.
In your writeup, report the accuracy of the trained model on the test set.
Remark. If you implement
Q naive Bayes the straightforward way, you will find that the
computed p(x|y) = i p(xi |y) often equals zero. This is because p(x|y), which is the
product of many numbers less than one, is a very small number. The standard computer
representation of real numbers cannot handle numbers that are too small, and instead
rounds them off to zero. (This is called “underflow.”) You’ll have to find a way to compute
Naive Bayes’ predicted class labels without explicitly representing very small numbers such
as p(x|y). [Hint: Think about using logarithms.]
Answer:
0.976702 with lowercasing 0.983871 after removing punctuations and numbers from the text in
the preprocessing step
1 Almeida, T.A., Gmez Hidalgo, J.M., Yamakami, A. Contributions to the Study of SMS Spam Filtering: New

Collection and Results. Proceedings of the 2011 ACM Symposium on Document Engineering (DOCENG’11),
Mountain View, CA, USA, 2011.
CS229 Problem Set #2 5

(c) [5 points] Intuitively, some tokens may be particularly indicative of an SMS being in a
particular class. We can try to get an informal sense of how indicative token i is for the
SPAM class by looking at:
 
p(xj = i|y = 1) P (token i|email is SPAM)
log = log .
p(xj = i|y = 0) P (token i|email is NOTSPAM)

Complete the get top five naive bayes words function within the provided code using
the above formula in order to obtain the 5 most indicative tokens.
Report the top five words in your writeup.
Answer:
I got these: claim won prize tone urgent! (with lowercasing) and : claim prize won tone
guaranteed (after removing punctuations and numbers)
(d) [2 points] Support vector machines (SVMs) are an alternative machine learning model that
we discussed in class. We have provided you an SVM implementation (using a radial basis
function (RBF) kernel) within src/spam/svm.py (You should not need to modify that
code).
One important part of training an SVM parameterized by an RBF kernel (a.k.a Gaussian
kernel) is choosing an appropriate kernel radius parameter.
Complete the compute best svm radius by writing code to compute the best SVM radius
which maximizes accuracy on the validation dataset. Report the best kernel radius you
obtained in the writeup.
Answer:
When using 0.1 as radius, accuracy: 0.971326.
CS229 Problem Set #2 6

3. [18 points] Constructing kernels

In class, we saw that by choosing a kernel K(x, z) = φ(x)T φ(z), we can implicitly map data to
a high dimensional space, and have a learning algorithm (e.g SVM or logistic regression) work
in that space. One way to generate kernels is to explicitly define the mapping φ to a higher
dimensional space, and then work out the corresponding K.
However in this question we are interested in direct construction of kernels. I.e., suppose we
have a function K(x, z) that we think gives an appropriate similarity measure for our learning
problem, and we are considering plugging K into the SVM as the kernel function. However for
K(x, z) to be a valid kernel, it must correspond to an inner product in some higher dimensional
space resulting from some feature mapping φ. Mercer’s theorem tells us that K(x, z) is a (Mercer)
kernel if and only if for any finite set {x(1) , . . . , x(n) }, the square matrix K ∈ Rn×n whose entries
are given by Kij = K(x(i) , x(j) ) is symmetric and positive semidefinite. You can find more
details about Mercer’s theorem in the notes, though the description above is sufficient for this
problem.
Now here comes the question: Let K1 , K2 be kernels over Rd × Rd , let a ∈ R+ be a positive
real number, let f : Rd 7→ R be a real-valued function, let φ : Rd → Rp be a function mapping
from Rd to Rp , let K3 be a kernel over Rp × Rp , and let p(x) a polynomial over x with positive
coefficients.
For each of the functions K below, state whether it is necessarily a kernel. If you think it is,
prove it; if you think it isn’t, give a counter-example.

(a) [1 points] K(x, z) = K1 (x, z) + K2 (x, z)

(b) [1 points] K(x, z) = K1 (x, z) − K2 (x, z)
(c) [1 points] K(x, z) = aK1 (x, z)
(d) [1 points] K(x, z) = −aK1 (x, z)
(e) [5 points] K(x, z) = K1 (x, z)K2 (x, z)
(f) [3 points] K(x, z) = f (x)f (z)
(g) [3 points] K(x, z) = K3 (φ(x), φ(z))
(h) [3 points] K(x, z) = p(K1 (x, z))

[Hint: For part (e), the answer is that K is indeed a kernel. You still have to prove it, though.
(This one may be harder than the rest.) This result may also be useful for another part of the
problem.]
Answer:

(a)Yes. According to notes, K is a valid kernel if z T Kz >= 0.

We know that z T (K1 )z >= 0 and z T (K2 )z >= 0,
thus z T Kz = z T (K1 )z + z T (K2 )z >= 0

(b)Not necessarily. While We know that z T (K1 )z >= 0 and z T (K2 )z >= 0,
we are not sure which one is larger, thus z T Kz >= 0 doesn’t necessarily hold.

(c)Yes. We know that z T (K1 )z >= 0 and a is a positive real number, thus z a ∗ T (K1 )z >= 0

(d)No. We know that z T (K1 )z >= 0 and a is a positive real number, thus z a ∗ T (K1 )z <= 0
CS229 Problem Set #2 7

(e)Yes. K1 (x, 2)K2 (x, 2)P= φ1 (x)> φ1 (z)φ2 (x) >

Pnφ2 (z)
m > >
= i=1 φ1 (x)i φ1 (z)i j=1 φ2 (x)j φ2 (z)j
Pm Pn
= i=1 j=1 [φi (x)i φi (x)j ] [φi (z)i φz (z)j ]
Pm Pn
= i=1 j=1 φ3 (x)ij φ3 (z)ij
=φ3 (x)> φ3 (z)
Pm Pm
(f)zT Kx = i=1,j=1 zi zj f (zi )f (zj ) = i=1 (zi f (zj ))
2
>= 0

(g)Yes. We know that K3 is a valid kernel. This is the same as question(c) if we set a to be 1.

(h)Yes. p(K1 (x, z)) can be view as a linear combination of K1 (x, z).
We know that the coefficients are positive, and that aK(x, z) is a valid kernel,
the combinations of valid kernels is also a valid kernel.
CS229 Problem Set #2 8

4. [15 points] Kernelizing the Perceptron

Let there be a binary classification problem with y ∈ {0, 1}. The perceptron uses hypotheses
of the form hθ (x) = g(θT x), where g(z) = sign(z) = 1 if z ≥ 0, 0 otherwise. In this problem
we will consider a stochastic gradient descent-like implementation of the perceptron algorithm
where each update to the parameters θ is made using only one training example. However, unlike
stochastic gradient descent, the perceptron algorithm will only make one pass through the entire
training set. The update rule for this version of the perceptron algorithm is given by
θ(i+1) := θ(i) + α(y (i+1) − hθ(i) (x(i+1) ))x(i+1)
where θ(i) is the value of the parameters after the algorithm has seen the first i training examples.
Prior to seeing any training examples, θ(0) is initialized to ~0.

(a) [3 points] Let K be a Mercer kernel corresponding to some very high-dimensional feature
mapping φ. Suppose φ is so high-dimensional (say, ∞-dimensional) that it’s infeasible to
ever represent φ(x) explicitly. Describe how you would apply the “kernel trick” to the
perceptron to make it work in the high-dimensional feature space φ, but without ever
explicitly computing φ(x).
[Note: You don’t have to worry about the intercept term. If you like, think of φ as having
the property that φ0 (x) = 1 so that this is taken care of.] Your description should specify:
i. [1 points] How you will (implicitly) represent the high-dimensional parameter vector
θ(i) , including how the initial value θ(0) = 0 is represented (note that θ(i) is now a
vector whose dimension is the same as the feature vectors φ(x));
ii. [1 points] How you will efficiently make a prediction on a new input x(i+1) . I.e., how
T
you will compute hθ(i) (x(i+1) ) = g(θ(i) φ(x(i+1) )), using your representation of θ(i) ;
and
iii. [1 points] How you will modify the update rule given above to perform an update to θ
on a new training example (x(i+1) , y (i+1) ); i.e., using the update rule corresponding to
the feature mapping φ:
θ(i+1) := θ(i) + α(y (i+1) − hθ(i) (x(i+1) ))φ(x(i+1) )
Answer:
i. K(θ, φ(x))
 >

(i)



ii. hθ x(i+1) = g θ(i) φ x(i+1) = g K θ(i) , φ x(i+1)






iii. θ(i+1) φ x(i+1) :=
θ(i) φ x(i+1) + α y (i+1)


− hθ(i) x(i+1) φ x(i+1) ) φ x(i+1)






K θ(π+1) , φ x(t+1) ) := K θ(i) , φ x(i+1) + α y (i+1) − g(K θ(i) , φ x(i+1) ))K φ x(i+1) , φ x(i+1) .
(b) [10 points] Implement your approach by completing the initial state, predict, and
update state methods of src/perceptron/perceptron.py.
We provide two kernels, a dot-product kernel and a radial basis function (RBF) kernel. Run
src/perceptron/perceptron.py to train kernelized perceptrons on src/perceptron/train.csv.
The code will then test the perceptron on src/perceptron/test.csv and save the resulting
predictions in the src/perceptron/ folder. Plots will also be saved in src/perceptron/.
Include the two plots (corresponding to each of the kernels) in your writeup, and indicate
which plot belongs to which kernel.
Answer:
CS229 Problem Set #2 9

Figure 2: dot
CS229 Problem Set #2 10

Figure 3: rbf
CS229 Problem Set #2 11

(c) [2 points]
One of the provided kernels performs extremely poorly in classifying the points. Which
kernel performs badly and why does it fail?
Answer:
Dot-product performs bad because i this case the boundary is a curve and the problem is not
linear.
CS229 Problem Set #2 12

5. [25 points] Neural Networks: MNIST image classification

In this problem, you will implement a simple neural network to classify grayscale images of
handwritten digits (0 - 9) from the MNIST dataset. The dataset contains 60,000 training images
and 10,000 testing images of handwritten digits, 0 - 9. Each image is 28×28 pixels in size, and
is generally represented as a flat vector of 784 numbers. It also includes labels for each example,
a number indicating the actual digit (0 - 9) handwritten in that image. A sample of a few such
images are shown below.

The data and starter code for this problem can be found in
• src/mnist/nn.py
• src/mnist/images train.csv
• src/mnist/labels train.csv
• src/mnist/images test.csv
• src/mnist/labels test.csv
The starter code splits the set of 60,000 training images and labels into a set of 50,000 examples
as the training set, and 10,000 examples for dev set.
To start, you will implement a neural network with a single hidden layer and cross entropy loss,
and train it with the provided data set. Use the sigmoid function as activation for the hidden
layer, and softmax function for the output layer. Recall that for a single example (x, y), the
cross entropy loss is:
K
X
CE(y, ŷ) = − yk log yˆk ,
k=1
K
where ŷ ∈ R is the vector of softmax outputs from the model for the training example x, and
y ∈ RK is the ground-truth vector for the training example x such that y = [0, ..., 0, 1, 0, ..., 0]>
contains a single 1 at the position of the correct class (also called a “one-hot” representation).
For n training examples, we average the cross entropy loss over the n examples.
n n K
1X 1 X X (i) (i)
J(W [1] , W [2] , b[1] , b[2] ) = CE(y (i) , ŷ (i) ) = − yk log ŷk .
n i=1 n i=1
k=1
CS229 Problem Set #2 13

The starter code already converts labels into one hot representations for you.
Instead of batch gradient descent or stochastic gradient descent, the common practice is to use
mini-batch gradient descent for deep learning tasks. In this case, the cost function is defined as
follows:

B
1 X
JM B = CE(y (i) , ŷ (i) )
B i=1
where B is the batch size, i.e. the number of training example in each mini-batch.

(a) [15 points]

Implement both forward-propagation and back-propagation for the above loss function.
Initialize the weights of the network by sampling values from a standard normal distribution.
Initialize the bias/intercept term to 0. Set the number of hidden units to be 300, and
learning rate to be 5. Set B = 1, 000 (mini batch size). This means that we train with
1,000 examples in each iteration. Therefore, for each epoch, we need 50 iterations to cover
the entire training data. The images are pre-shuffled. So you don’t need to randomly
sample the data, and can just create mini-batches sequentially.
Train the model with mini-batch gradient descent as described above. Run the training for
30 epochs. At the end of each epoch, calculate the value of loss function averaged over the
entire training set, and plot it (y-axis) against the number of epochs (x-axis). In the same
image, plot the value of the loss function averaged over the dev set, and plot it against the
number of epochs.
Similarly, in a new image, plot the accuracy (on y-axis) over the training set, measured as
the fraction of correctly classified examples, versus the number of epochs (x-axis). In the
same image, also plot the accuracy over the dev set versus number of epochs.
Submit the two plots (one for loss vs epoch, another for accuracy vs epoch) in
your writeup.
Also, at the end of 30 epochs, save the learnt parameters (i.e all the weights and biases) into
a file, so that next time you can directly initialize the parameters with these values from
the file, rather than re-training all over. You do NOT need to submit these parameters.
Hint: Be sure to vectorize your code as much as possible! Training can be very slow
otherwise.
Answer:
CS229 Problem Set #2 14

(b) [7 points] Now add a regularization term to your cross entropy loss. The loss function will
become !
B
1 X  
JM B = CE(y , ŷ ) + λ ||W [1] ||2 + ||W [2] ||2
(i) (i)
B i=1

Be careful not to regularize the bias/intercept term. Set λ to be 0.0001. Implement the
regularized version and plot the same figures as part (a). Be careful NOT to include the
regularization term to measure the loss value for plotting (i.e., regularization should only
be used for gradient calculation for the purpose of training).
Submit the two new plots obtained with regularized training (i.e loss (without
regularization term) vs epoch, and accuracy vs epoch) in your writeup.
Compare the plots obtained from the regularized model with the plots obtained from the
non-regularized model, and summarize your observations in a couple of sentences.
As in the previous part, save the learnt parameters (weights and biases) into a different file
so that we can initialize from them next time.
Answer:
CS229 Problem Set #2 15

With regularization, the training accuracy seems similar to without regularization, while the test
accuracy is slightly higher than without regularization, especially after the first few epochs.
(c) [3 points] All this while you should have stayed away from the test data completely. Now
that you have convinced yourself that the model is working as expected (i.e, the observations
you made in the previous part matches what you learnt in class about regularization), it is
finally time to measure the model performance on the test set. Once we measure the test
set performance, we report it (whatever value it may be), and NOT go back and refine the
model any further.
Initialize your model from the parameters saved in part (a) (i.e, the non-regularized model),
and evaluate the model performance on the test data. Repeat this using the parameters
saved in part (b) (i.e, the regularized model).
Report your test accuracy for both regularized model and non-regularized model.
Answer: regularized: 0.9320, non regularized: 0.9287
CS229 Problem Set #2 16

6. [20 points] Bayesian Interpretation of Regularization

Background: In Bayesian statistics, almost every quantity is a random variable, which can
either be observed or unobserved. For instance, parameters θ are generally unobserved random
variables, and data x and y are observed random variables. The joint distribution of all the
random variables is also called the model (e.g., p(x, y, θ)). Every unknown quantity can be esti-
mated by conditioning the model on all the observed quantities. Such a conditional distribution
over the unobserved random variables, conditioned on the observed random variables, is called
the posterior distribution. For instance p(θ|x, y) is the posterior distribution in the machine
learning context. A consequence of this approach is that we are required to endow our model
parameters, i.e., p(θ), with a prior distribution. The prior probabilities are to be assigned before
we see the data—they capture our prior beliefs of what the model parameters might be before
observing any evidence.
In the purest Bayesian interpretation, we are required to keep the entire posterior distribu-
tion over the parameters all the way until prediction, to come up with the posterior predictive
distribution, and the final prediction will be the expected value of the posterior predictive dis-
tribution. However in most situations, this is computationally very expensive, and we settle for
a compromise that is less pure (in the Bayesian sense).
The compromise is to estimate a point value of the parameters (instead of the full distribution)
which is the mode of the posterior distribution. Estimating the mode of the posterior distribution
is also called maximum a posteriori estimation (MAP). That is,

θMAP = arg max p(θ|x, y).

θ

Compare this to the maximum likelihood estimation (MLE) we have seen previously:

θMLE = arg max p(y|x, θ).

θ

In this problem, we explore the connection between MAP estimation, and common regularization
techniques that are applied with MLE estimation. In particular, you will show how the choice
of prior distribution over θ (e.g., Gaussian or Laplace prior) is equivalent to different kinds of
regularization (e.g., L2 , or L1 regularization). To show this, we shall proceed step by step,
showing intermediate steps.

(a) [3 points] Show that θMAP = argmaxθ p(y|x, θ)p(θ) if we assume that p(θ) = p(θ|x). The
assumption that p(θ) = p(θ|x) will be valid for models such as linear regression where the
input x are not explicitly modeled by θ. (Note that this means x and θ are marginally
independent, but not conditionally independent when y is given.)
Answer:
θMAP = argmaxθ p(θ|x, y) = argmaxθ p(x, y|θ)P (θ) = argmaxθ p(x,y,θ)
P (θ) ∗ P (θ)
p(θ,x,y) p(θ,x,y)
On the other side, argmaxθ p(y|x, θ)∗P (θ) = argmaxθ P (x,θ) ∗P (θ) = argmaxθ P (θ) ∗P (θ)
Thus, the two are equivelent.
(b) [5 points] Recall that L2 regularization penalizes the L2 norm of the parameters while
minimizing the loss (i.e., negative log likelihood in case of probabilistic models). Now we
will show that MAP estimation with a zero-mean Gaussian prior over θ, specifically θ ∼
N (0, η 2 I), is equivalent to applying L2 regularization with MLE estimation. Specifically,
show that
θMAP = arg min − log p(y|x, θ) + λ||θ||22 .
θ
CS229 Problem Set #2 17

Also, what is the value of λ?

Answer:
θM AP = argminθ (− log [P (y|x, θ) P (θ)]
=arg minθ − log P (y|x, θ) − log P (θ)
θ 2
=argminθ − log P (y|x, θ) − log √ 1
2πτ 2 I
e− 2τ 2 I
θ2
=arg minθ − log P (y|x, θ) + 2τ 2 I
1 2
=arg minθ − log P (y|x, θ) + 2τ 2 I kθk2
1
λ= 2τ 2 I

(c) [7 points] Now consider a specific instance, a linear regression model given by y = θT x + 
where  ∼ N (0, σ 2 ). Assume that the random noise (i) is independent for every training
example x(i) . Like before, assume a Gaussian prior on this model such that θ ∼ N (0, η 2 I).
For notation, let X be the design matrix of all the training example inputs where each row
vector is one example input, and ~y be the column vector of all the example outputs.
Come up with a closed form expression for θMAP .
Answer:    
2
θ2
P(θ) = √ 12 exp − (θ−µ 0)
2η 2 I ∝ exp − 2η 2 I
2η Iπ
 T 2

(ŷ−θ x)
 2

P (θ|ŷ, x) = P (ŷ|x, θ)P (θ) ∝ exp − 2σ2 exp − 2ηθ 2 I
θM AP = argminθ [− logP (ŷ|x, θ)P (θ)]

2
(ŷ−θT x)
 2

= argminθ [] − log[exp − 2σ2 exp − 2ηθ 2 I ]]
2
= argminθ [ 2σ1 2 kŷ − Xθk2 + 1 2
2η 2 I kθk2 ]
(d) [5 points] Next, consider the Laplace distribution, whose density is given by

 
1 |z − µ|
fL (z|µ, b) = exp − .
2b b

As before, consider a linear regression model given by y = xT θ +  where  ∼ N (0, σ 2 ).

Assume a Laplace prior on this model, where each parameter θi is marginally independent,
and is distributed as θi ∼ L(0, b).
Show that θMAP in this case is equivalent to the solution of linear regression with L1
regularization, whose loss is specified as

J(θ) = ||Xθ − ~y ||22 + γ||θ||1

Also, what is the value of γ?
Note: A closed form solution for linear regression problem with L1 regularization does not
exist. To optimize this, we use gradient descent with a random initialization and solve it
numerically.
Answer:    
P(θ) = 2b 1
exp − |θ|b ∝ exp − |θ|
b
 2

(ŷ−θT x)
 
P (θ|ŷ, x) = P (ŷ|x, θ)P (θ) ∝ exp − 2σ2 exp − |θ|
b

θM AP = argminθ [− log P (ŷ|x, θ)P (θ)]

CS229 Problem Set #2 18

 2

(ŷ−θT x)
 
= argminθ [] − log[exp − 2σ2 exp − |θ|
b ]]
2
= argminθ [ 2σ1 2 kŷ − Xθk2 + 1b kθk1 ]
2
J(θ) = kXθ − ~y k22 + 2σb kθk1
2
γ = 2σb

Remark: Linear regression with L2 regularization is also commonly called Ridge regression, and
when L1 regularization is employed, is commonly called Lasso regression. These regularizations
can be applied to any Generalized Linear models just as above (by replacing log p(y|x, θ) with
the appropriate family likelihood). Regularization techniques of the above type are also called
weight decay, and shrinkage. The Gaussian and Laplace priors encourage the parameter values
to be closer to their mean (i.e., zero), which results in the shrinkage effect.
Remark: Lasso regression (i.e., L1 regularization) is known to result in sparse parameters,
where most of the parameter values are zero, with only some of them non-zero.