Machine learning exam questions, ML solved quiz questions, Machine Learning TRUE or FALSE questions
Machine Learning TRUE / FALSE Questions  SET 16
1. Using the kernel trick, one can get nonlinear decision boundaries using algorithms designed originally for linear models.
(a) TRUE (b) FALSE
Answer: TRUE Kernel trick solves the nonlinear decision boundary problem. Kernel trick is simply increasing the number of dimensions. It is to make the nonlinear decision boundary in lower dimensional space as a linear decision boundary, in higher dimensional space. In simple words, Kernel trick makes the nonlinear decision boundary to linear (in higher dimensional space). This is helpful in SVM. SVM works well if the data points are linearly separable. In nonlinear boundary case, it is difficult for SVM to classify. In this case, we can use kernel trick to convert nonlinear boundary to linear. 
2. Zero correlation between any two random variables implies that the two random variables are independent.
(a) TRUE (b) FALSE
Answer: FALSE If ρ(X, Y)=0 we say that X and Y are “uncorrelated”. If two variables are independent, then their correlation will be 0. A correlation of 0 does not imply independence. If X and Y are uncorrelated, then they can still be dependent. Example: Refer here http://mathforum.org/library/drmath/view/64808.html 
3. In linear SVMs, the optimal weight vector w is a linear combination of training data points.
(a) TRUE (b) FALSE
Answer: TRUE The optimal weight vector w is a linear combination of the training data points including training inputs and the training outputs, x_{i} (data point) and y_{i} (class).

4. The maximum likelihood estimate for the variance of a univariate Gaussian is unbiased.
(a) TRUE (b) FALSE
Answer: FALSE The maximum likelihood estimate (MLE) for the variance of a univariate Gaussian is biased. But we can construct an unbiased estimator based on MLE. The MLE estimator is a biased estimator of the population variance and it introduces a downward bias (underestimating the parameter). The size of the bias is proportional to population variance, and it will decrease as the sample size gets larger. MLE Maximum Likelihood Estimation (MLE) is a method of estimating the parameters of a statistical model. It is widely used in Machine Learning algorithm, as it is intuitive and easy to form given the data. The basic idea underlying MLE is to represent the likelihood over the data w.r.t the model parameters, then find the values of the parameters so that the likelihood is maximized. 
5. With a nonlinearlyseparable dataset that contains some extra “noise” data points, using an SVM with slack variables to create a soft margin classifier, and a small value for the penalty parameter, C, that controls how much to penalize misclassified points, will often reduce overfitting the training data.
(a) TRUE (b) FALSE
Answer: TRUE Small C means the penalty for misclassifying a few points will be small and therefore we are more likely to maximize the margin between most of the points while misclassifying a few points including the noise points.

*********************