The author took the machine learning exam from 13:00 to 15:00 on October 30, 2022. Since machine learning is not a credit course this year, the overall difficulty has been reduced.
Since scratch paper was collected during the exam, all the following questions are based on memories after the exam and are not necessarily accurate and are for reference only.
Since scratch paper was collected during the exam, all the following questions are based on memories after the exam and are not necessarily accurate and are for reference only.
Since scratch paper was collected during the exam, all the following questions are based on memories after the exam and are not necessarily accurate and are for reference only.
-
What is supervised learning? What is unsupervised learning? Each category gives the names of 3 algorithms.
-
Explain entropy, conditional entropy, and mutual information. During the generation process of the decision tree, for a certain node, given attributes X1 and X2, the classification results are as follows. Which attribute is better to choose? Please explain why.
-
The steps of the K-Means algorithm; the optimization function of the K-Means algorithm; why can the key steps of the K-Means algorithm optimize the optimization function?
-
What is overfitting? Please explain with experiments you have done; and give two methods to solve overfitting.
-
For Bayesian classification, assuming there are two classes and the class conditional distribution is Gaussian distribution, what conditions does Gaussian distribution satisfy to make the decision surface of Bayesian classification a linear decision surface?
-
Try to analyze the similarities and differences between SVM and logistic regression in terms of loss function, decision plane, etc.;
-
What are your thoughts on learning machine learning?