Home / Past papers / MLF End Term / this paper
defined as
. For what values of
is
convex 




Which of the following is/are true 
is Hermitian.
is unitary.
is positive semi-definite.
is positive definite.
is invertible.
or not
by using some classification models. Therefore, the team defines three binary features
where,
= Presence of suspicious keywords (1 = present, 0 = absent)•
= Whether the sender is unknown (1 = unknown, 0 = known)•
= Whether the email contains a link (1 = yes, 0 = no)• The team collects the following dataset from previously classified emails:
The model makes a prediction using the rule :
and
Compute the average loss on this dataset. Enter your answer correct to one decimal place.
be a continuous random variable with the probability density function:
What is the value of 
is a
matrix whose eigenvalues are
. Consider the set
as defined below:
Find
.
with radius
that is closest to the point
. If this distance is
, enter
as your answer. Your answer should be an integer.
follows an exponential distribution with mean 5 minutes. Based on the above data, answer the given subquestions.
such that 95% of customers are served within
minutes.



and
minutes 




, where
represents the mean of the random variable
. Enter your answer correct to two decimal places.
Answer the sub-questions assuming that we take a very small step in the direction of interest.
which of the following is/are descent directions for the objective function 




which of the following is/are feasible directions 




a potential candidate at which the problem could have a local minimum 
and
, where
Based on the above data, answer the given subquestions.








.



The company applies the gradient descent algorithm to minimize this function, starting with an initial guess of
and using a step size (learning rate) of
, where
Based on the above data, answer the given subquestions.
after the first iteration of the gradient descent algorithm.



after the second iteration, i.e.,
.
and the price vector
(in hundreds of thousands of rupees) are:
The objective is to find the weight vector
that minimizes the sum of squared errors given by the loss function:
If
is the value that minimizes
, then which of the following is
?



. Find the directional derivative of
at the point
in the direction of unit vector along
.



. Which of the following points belongs to the convex hull?



is a square invertible matrix, which of the following is/are true?
is equal to its column space.
is equal to its left nullspace.
is equal to its row space.
is equal to its row space.
denote number of defective items produced in a day. Using Markov's inequality, the analyst finds an upper bound of
on the probability that at least
items are defective. Calculate the mean of
.
,
and
, each having three points as shown below. The figure is drawn to scale. PCA is run on both these datasets separately.
and
are the covariance matrices of the two datasets respectively. If
is the eigenvector corresponding to the largest eigenvalue of
and
is the eigenvector corresponding to the smallest eigenvalue of
, find the absolute value of
. Both eigenvectors have unit norm. Your answer should be an integer.
. The time spent (in minutes) from a random sample of 5 customers are as follows:
Find the maximum likelihood estimate of
for the given sample. Enter the answer correct to two decimal places.
denote the distance from the center to the point where the dart hits the board. Suppose the PDF of
is
Calculate the value of
. Enter the answer correct to two decimal places.
with radius
that is closest to the point
. If this distance is
, enter
as your answer. Your answer should be an integer.
If
corresponds to the global minimum, find
. Enter your answer correct to one decimal place.
and
, where
Based on the above data, answer the given subquestions.








.



The company applies the gradient descent algorithm to minimize this function, starting with an initial guess of
and using a step size (learning rate) of
, where
Based on the above data, answer the given subquestions.
after the first iteration of the gradient descent algorithm.



after the second iteration, i.e.,
.