Nirma University Institute of Technology Semester End Examination (IR) / Supplementary Examination, May - 2022 B, Tech. in Computer Science and Engineering, Semester-V 2C$501 Machine Learning Roll / Supervisor's initial Exam No. with date Time: 2:00 Hours : Max, Marks: 50 Instructions: 4. Attempt all questions. 2. Figures to right indicate full marks, 3, Draw neat sketches wherever necessary. 4. Make suitable assumptions wherever necessary. Qa Answer the following questions: [10] @) Consider the two groups of data points (circles and squares) as shown in [05] Figure below. Use SVM kernel function, using the transformation function ®, to find hyper plane equation. 6-11 +(q-1) ae idty22 6-12 + (K-) “ 8 ( } otherwise vy) ‘Transform the points circle and square and plot it in the new feature space. {b) Define the terms [05] i) Indicators of overfitting and underfitting, ii) Regularization and its applications Q.2 Answer the following questions: . [15] (a) Which ML algorithm will you use in following situation explain with proper [05] justification? (i) Predicting will it rain or not based on wind speed, temperature, humidity in air. (ii) Predicting price of the house based on size, age of the house, distance from the main city. (>) Given data as in table below. Assuming that the MO= (4 (no intercept), [05] also the cost function is given as KO = 1/2myzqr—h(0 J'2_. Plot J(@) against 0 for 0 (0,1,2,3.4) x = a ea 21s cae 7s cae aa Saee|ao Page 1 of 3(b) te) fa) (a) (b) 2CS501, Machine Learning OR Define the term decision boundary, explain the same using appropriate example and demonstration. (i) Take an appropriate example for reinforcement learning and define the [05] components based on MDP. (i) How is reinforcement learning paradigm different than supervised and unsupervised learning? Rind the root node using information en for the dataset as below [05] Animal | Warm- | Feathers | Fur | Swims | Lays blooded Eggs Ostrich Yes Yes No | No Yes Crocodile _| No No No | Yes Yes Raven Yes Yes No | No Yes Albatross_| Yes Yes No | No Yes Dolphin | Yes No No | Yes No Koala Yes No Yes | No No ‘Answer the following questions: [15] What are different loss functions? How to choose a loss function given a [05] problem? Explain the same with appropriate example. OR Write differences between (05) a. Self-organizing Maps and Artificial Neural networks b. Gradient descent and Normal equation for regression ‘The network shown in figure below , when properly trained should respond [10] with 0) (0.95 01} = [0.05 To the argument input pattern fe The network weights have been initialized as shown in the figure. Analyze a single feed forward and back propagation step for the initialized network by doing the following (a) Find weight matrices. (b) Calculate output at each neuron. Page 2 of 3Q4 (a) (b) 2CS501, Machine Leaning, (c} Calculate error signals at each neurons For the computation assume finet)=[1#exp(-net]}! and 9=1 Answer the following question: [10] Assume training tuples: (85, 85, No), (71, 80, No}, (75, 50, Yes), (60, [05] 80, Yes), (72, 3, Yes). The last value in each tuple is the class to which the tuple belongs. Using distance/similarity-weighted 3-NN, predict the class of the tuple (72, 75). Assume cosine similarity as the similarity/distance measure. Use Naive Bayes Algorith for predicting the class for X = (outlook: [05] sunny, temperature: mild, humidity: normal, windy: false) Outlook Temperature Humidity Windy Play? sy Te Tig FALSE 7 sy a gh TRUE 0 overcast het ih Fase ve ‘shy mis igh Fase ve any cool emai FALSE we wey cool remsi TRUE 70 overca too toms! TRUE vs sunay i igh Fase 0 amy coo romsi FALSE. ye rary oa romal “FALSE. yes soy os reomal TRUE 8 overcast aa igh TRUE ve vice hot memel FALSE yes oy a igh TRUE 2 Page 3 of 3