Machine Learning Set 3 MCQs

Q1 | In multiclass classification number of classes must be

less than two

equals to two

greater than two

option 1 and option 2

Q2 | Which of the following can only be used when training data are linearlyseparable?

linear hard-margin svm

linear logistic regression

linear soft margin svm

the centroid method

Q3 | Impact of high variance on the training set ?

overfitting

underfitting

both underfitting & overfitting

depents upon the dataset

Q4 | What do you mean by a hard margin?

the svm allows very low error in classification

the svm allows high amount of error in classification

both 1 & 2

none of the above

Q5 | The effectiveness of an SVM depends upon:

selection of kernel

kernel parameters

soft margin parameter c

all of the above

Q6 | What are support vectors?

all the examples that have a non-zero weight ??k in a svm

the only examples necessary to compute f(x) in an svm.

all of the above

none of the above

Q7 | A perceptron adds up all the weighted inputs it receives, and if it exceeds a certain value, it outputs a 1, otherwise it just outputs a 0.

true

false

sometimes – it can also output intermediate values as well

can’t say

Q8 | What is the purpose of the Kernel Trick?

to transform the data from nonlinearly separable to linearly separable

to transform the problem from regression to classification

to transform the problem from supervised to unsupervised learning.

all of the above

Q9 | Which of the following can only be used when training data are linearlyseparable?

linear hard-margin svm

linear logistic regression

linear soft margin svm

parzen windows

Q10 | The firing rate of a neuron

determines how strongly the dendrites of theneuron stimulate axons of neighboring neurons

is more analogous to the output of a unit in aneural net than the output voltage of the neuron

only changes very slowly, taking a period ofseveral seconds to make large adjustments

can sometimes exceed 30,000 action potentialsper second

Q11 | Which of the following evaluation metrics can not be applied in case of logistic regression output to compare with target?

auc-roc

accuracy

logloss

mean-squared-error

Q12 | The cost parameter in the SVM means:

the number of cross-validations to be made

the kernel to be used

the tradeoff between misclassification and simplicity of the model

none of the above

Q13 | The kernel trick

can be applied to every classification algorithm

is commonly used for dimensionality reduction

changes ridge regression so we solve a d ?? dlinear system instead of an n ?? n system, given nsample points with d features

exploits the fact that in many learning algorithms, the weights can be written as a linearcombination of input points

Q14 | How does the bias-variance decomposition of a ridge regression estimator compare with that of ordinaryleast squares regression?

ridge has larger bias, larger variance

ridge has smaller bias, larger variance

ridge has larger bias, smaller variance

ridge has smaller bias, smaller variance

Q15 | Which of the following are real world applications of the SVM?

text and hypertext categorization

image classification

clustering of news articles

all of the above

Q16 | How can SVM be classified?

it is a model trained using unsupervised learning. it can be used for classification and regression.

it is a model trained using unsupervised learning. it can be used for classification but not for regression.

it is a model trained using supervised learning. it can be used for classification and regression.

t is a model trained using unsupervised learning. it can be used for classification but not for regression.

Q17 | Which of the following can help to reduce overfitting in an SVM classifier?

use of slack variables

high-degree polynomial features

normalizing the data

setting a very low learning rate

Q18 | Suppose you have trained an SVM with linear decision boundary after training SVM, you correctly infer that your SVM model is under fitting. Which of the following is best option would you more likely to consider iterating SVM next time?

you want to increase your data points

you want to decrease your data points

you will try to calculate more variables

you will try to reduce the features

Q19 | What is/are true about kernel in SVM? 1. Kernel function map low dimensional data to high dimensional space2. It’s a similarity function

1

2

1 and 2

none of these

Q20 | You trained a binary classifier model which gives very high accuracy on the training data, but much lower accuracy on validation data. Which is false.

this is an instance of overfitting

this is an instance of underfitting

the training was not well regularized

the training and testing examples are sampled from different distributions

Q21 | Suppose your model is demonstrating high variance across the different training sets. Which of the following is NOT valid way to try and reduce the variance?

increase the amount of traning data in each traning set

improve the optimization algorithm being used for error minimization.

decrease the model complexity

reduce the noise in the training data

Q22 | Suppose you are using RBF kernel in SVM with high Gamma value. What does this signify?

the model would consider even far away points from hyperplane for modeling

the model would consider only the points close to the hyperplane for modeling

the model would not be affected by distance of points from hyperplane for modeling

none of the above

Q23 | We usually use feature normalization before using the Gaussian kernel in SVM. What is true about feature normalization? 1. We do feature normalization so that new feature will dominate other2. Some times, feature normalization is not feasible in case of categorical variables3. Feature normalization always helps when we use Gaussian kernel in SVM

1

1 and 2

1 and 3

2 and 3

Q24 | Wrapper methods are hyper-parameter selection methods that

should be used whenever possible because they are computationally efficient

should be avoided unless there are no other options because they are always prone to overfitting.

are useful mainly when the learning machines are “black boxes”

should be avoided altogether.

Q25 | Which of the following methods can not achieve zero training error on any linearly separable dataset?

decision tree

15-nearest neighbors

hard-margin svm

perceptron