# 机器学习代写｜CSC311 Introduction to Machine Learning Midterm Test

1. (16 points total) Multiple choice.

In each question below, indicate which answer is best. Do not give an explanation. (2 points
each)

(a) Determining whether a piece of text contains positive or negative sentiment is an example
of…

i. A regression problem
ii. A k-Nearest Neighbour problem
iii. A classification problem
iv. A text generation problem
v. A naive Bayes problem

(b) Which of the following statements about the training, validation and test sets is false?

i. The training set should be used to estimate model parameters.
ii. The validation set should be used to estimate hyperparameters.
iii. The test set should be used to decide which model will likely perform the best when
deployed.
iv. All of the above are true
v. All of the above are false

(c) Which of the following is a weakness of the k-nearest neighbour model?

i. It takes a long time to train the model
ii. It takes a long time to make predictions
iii. The model requires us to store the training set in order to make predictions
iv. Both ii and iii are true.

(d) Which of the following loss functions cannot be optimized using gradient descent?

i. 0-1 loss
ii. cross entropy loss
iii. square error loss
iv. L(y; t) = jy − tj + (y − t)2
v. Neither i or iv can be optimized using gradient descent

(e) Which of the following is a possible output of the softmax function-call softmax([a; b; c]),
where a, b and c are real numbers?

i. 0
ii. [1, 2, 3]
iii. [0.5, 0.5, 0.1]
iv. [0.5, 0.4, 0.1]
v. [1, 0, 1]
vi. None of the above

(f) Which of the following is true about the conditional independence assumption used in
Naive Bayes?

i. Conditional independence is another way of expressing that two features x1; x2 are
statistically independent
ii. Conditional independence is another way of expressing that two data points are
independent and identically distributed
iii. Conditional independence does not usually hold in a real data set, but is a simplifying
assumption.
iv. All of the above are true

E-mail: itcsdx@outlook.com  微信:itcsdx