COSC2673/COSC2793 | Machine Learning
Tutorial | Week 03
Tutorial Questions
1. Define a Linear Regression Problem. What is: (a) The task?
(b) The general form of the hypothesis? (c) The experience?
2. What assumptions are made by using a linear regression model? What are the strengths and limitations of linear regression?
3. Describe the loss function for linear regression. Give an intuitive explanation of the term(s) of the loss function. What happens if the loss function is 0?
4. This was a previous exam question:
Consider the following training data set, for a univariate target function, y = f(x1), over real-valued numbers. The attribute is, x1, and the output is, y.
y x1 2 0.5 4.1 1.0 4.5 2.0 6.4 2.5 7.3 3.0
Use one (1) iteration of Linear Regression Gradient Descent to calculate values for the parameter(s) of a linear regression model. Set the initial value of each of the linear regression parameter(s) to one, for example ¦È0 = 1, and use a learning rate of ¦Á = 0.1.
5. Derive the gradient descent update rule for univariate linear regression.
1