Name+Email ID
CSCI 5521: Machine Learning Fundamentals (Fall 2021)
Quiz 1 (Thurs, Sep 30)
Due on Gradescope at 02:00 PM, Friday, Oct 1
Instructions:
• This quiz has 3 questions, 30 points, on 1 page.
• Please write your name & ID on your submission pages.
1. (8 points) Supervised learning has a wide range of applications. For example, pedestrian detection
is used in self-driving cars to detect “pedestrians” on road. Please model the question as a two-class
classification task and answer the following questions.
(a) What are the labels in this two-class classification task?
(b) Name one relevant feature to this two-class classification task. (Any reasonable features are
acceptable).
2. (8 points) What could we do to reduce overfitting in a regression model? Select all the option(s)
that apply.
(a) Add new training data and keep test data the same.
(b) Add new test data and keep training data the same.
(c) Decrease the number of parameters.
(d) Increase the complexity of the model.
3. (14 points) The conditional probability density functions of two classes C1 and C2 are shown in
the figure below, with P (x|C1) ∼ N (0, 1) and P (x|C2) ∼ N (1, 1).
(a) Assuming the priors are equal, predict which class (C1 or C2) the data point x = 0.7 (illus-
trated with the green dot) belongs to. Briefly explain why.
(b) What if the priors are P (C1) = 0.8 and P (C2) = 0.2, respectively? (Note: High-level
explanation is good, but you can use formulations if it helps explain).
Page 1