Question 1 : Why SVM’s are more accurate than logistic regression?
- SVM gives more weightage to wrongly classified data points.
- SVM gives more weightages to data points which are correctively classified .
- SVM uses all the data points assuming a probabilistic model.
- SVM uses concept of large margin seperator and for non linearity it uses kernel functions
Question 2 : What is true about the discount factor in reinforcement learning?
- discount factor should be greater than 1
- discount factor should always be negative
- discount factor should be in range of 0 and 1
- discount factor can be any real number
Question 3 : Which algorithm is used for performing probabilistic reasoning on temporal data?
- Hill-climbing search
- Hidden markov model
- Naïve Method
- Support Vector Machine
Question 4 : Choose correct applications of reinforcement learning?
- Aircraft Control
- Sentimental analysis
- House price prediction
- Spam Email Filtering
Question 5 : Support Vector Machine(SVM) can be used for both classification or regression challenges.Which kind of learning technique SVM uses?
- supervised
- unsupervised
- reinforced
- clustered
Question 6 : What are two steps of pruning in decision tree ?
- Pessimistic pruning and Optimistic pruning
- Postpruning and Prepruning
- Cost complexity pruning and time complexity pruning
- High pruning and low pruning
Question 7 : Various ____ methods and techniques are used for calculation of the outliers.
- distance calculation
- prediction
- optimization
- integration
Question 8 : Which of the following is a clustering algorithm in machine learning?
- Expectation Maximization
- CART
- Gaussian Naïve Bayes
- Apriori
Question 9 : What is the approach of basic algorithm for decision tree induction?
- Greedy
- Top Down
- Procedural
- Step by Step
Question 10 : Support Vector Machine(SVM) performs well in _____ dimension spaces.
- high
- low
- wide
- single
Question 11 : You ran gardient descent for 20 iterations with learning rate=0.2 and compute cost for each iteration.You observe that cost decreases after each iteration .Based on this which conclusion is more suitable.
- Try smaller values of learning rate like 0.01
- 0.2 is effective choice of learning rate.
- Try larger values of learning rate like 1.
- Try any negetive value for learning rate
Question 12 : What are support vectors?
- These are the datapoints which help the SVM to generate optimal hyperplane.
- It is an intermediate vector generated during calculation of optimal hyperplane
- In SVM all the data points are called support vectors.
- This are predefined vectors used in calculating hyperplane
Question 13 : Which of the following problems can be solved by supervised learning too? Assume appropriate dataset is available.
- From a large collection of spam emails, discover if there are sub types of spam emails.
- Given data on how 1000 medical patients respond to an experimental medicine , discover whether there are different categories of patients in terms of how they respond to , and if so what are these categories
- Given a large dataset of medical records from patients suffering from heart disease, try to learn whether there might be different groups of patients for which customised treatment is required
- Given genetic (DNA) data from a person, predict the odds of the person developing diabetes over the next 10 years
Question 14 : Principal component analysis(PCA) is used for___.
- Dimensionality Enhancement
- LU Decomposition
- QR Decomposition
- Dimensionality Reduction
Question 15 : ___________ phenomenon refers that a model is neither trained on training data nor generalized properly on new data.
- good fitting
- overfitting
- moderate fitting
- underfitting
Question 16 : In Logistic regression predicted value of Y lies within _____ range.
- 0 to 1
- 0 to -∞
- -∞ to +∞
- -1 to 1
Question 17 : Choose the correct tree based learner.
- Rule based
- Hidden markov model
- Bayesian classifier
- CART
Question 18 : In principal component analysis ,if eigenvalues are equal.What does it mean?
- PCA will perform outstandingly
- PCA will perform badly
- No effect
- Model will be unstable
Question 19 : For a trained logistic classifer given a sample x,it gives prediction as 0.8.This means that___.
- P(Y=0|x)=0.8
- P(Y=1|x)=0.8
- P(Y=0|x)=0.2
- P(Y=1|x)=0.2
Question 20 : K-fold cross-validation is____.
- linear in K
- quadratic in K
- cubic in K
- exponential in K
Question 21 : Consider a point that is correctly classified and distant from the decision boundary. Which of the following methods will be unaffected by this point?
- Nearest neighbor
- Support Vector Machine
- Logistic regression
- Linear regression
Question 22 : You are training an RBF SVM with the following parameters: C (slack penalty) and γ = 1/2σ 2 (where σ 2 is the variance of the RBF kernel). How should you tweak the parameters to reduce overfitting?
- Increase C and/or reduce γ
- Reduce C and/or increase γ
- Reduce C and/or reduce γ
- Reduce C only (γ has no predictable effect on overfitting)
Question 23 : Machine Learning comes under which of the following domain?
- Artificial Intelligence
- Network Security
- Engineeering sciences
- System programming
Question 24 : Choose the reason for pruning a Decision Tree?
- To save computing time during testing
- To avoid overfitting the training set
- To save space for storing the Decision Tree
- To make the training set error smaller
Question 25 : While comparing reinforcement learning and supervised learning, which of the following statement is true?
- Both in reinforcement and supervised learning decisions are taken sequentially
- Supervised learning is best suited where human interaction is prevalant wheareas reinforcement learning is best suited for sofware systems.
- Reinforcement learning works by interacting with environment wheareas supervised learning works on sample data
- Both in reinforcement and supervised learning decisions taken at one time step is independent with respect to previous timestep.
Question 26 : Neural networks:
- Optimize a convex objective function
- Can use a mix of different activation functions
- are not suitable for learning.
- Can only be trained with stochastic gradient descent
Question 27 : Which of the following is not a clustering algorithm?
- EM-Algorithm
- K-means clustering
- Radial Basis Function
- Decision Tree
Question 28 : Principal component analysis is a technique for performing
- Dimensionality reduction
- Pruning
- Aggregation
- Sampling
Question 29 : The process of obtaining best result under given constraints is called as
- Optimization
- Generalization
- Summation
- Regularization
Question 30 : Below are the 8 actual values of the target variable in the train file.[0,0,0,1,1,1,1,1]What is the entropy of the target variable?
- -(5/8 log(5/8) + 3/8 log(3/8))
- 5/8 log(5/8) + 3/8 log(3/8)
- 3/8 log(5/8) + 5/8 log(3/8)
- 5/8 log(3/8) – 3/8 log(5/8)