Thanks to visit codestin.com
Credit goes to www.scribd.com

0% found this document useful (0 votes)
75 views9 pages

Pattern Recognition Pyq

Uploaded by

Krupa Janani
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
75 views9 pages

Pattern Recognition Pyq

Uploaded by

Krupa Janani
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 9
V SEMESTER ASSESSMENT -1l Date: 07-12-2622 Mayinnim Marks: Course Outcomes On completion o: 1. Implement basic pattern classifier 2. Have knowledge about the working 3. Have knowledge about functionality of classifiers. 4, Pereeive the recent advancement in pattern recognition: 5. Apply SVM and HMM algorithms for real time a course, 1 students will be able to: Igorithms. lications. 6. Implement advanced methodologies over image processing applications. inciple of unsupervised algorithms. Part A 1 T QNo. Questions Marks | CO | BL 1 | Write the advantages and the limitations of clustering . algorithms in a 2 | Find the City block distance and Chebyshev distance | | between the data points X and Y Ce 4 X=[10, 20, 15, 10, 5] Y=[12, 24, 18, 8, 7] \ 3 _ | Define Grid-based clustering. Name two Grid-based | clustering approaches. (an 4 [Define Markov process. Given that the weather on day I (t= 1) is sunny (S), what is the probability that the weather for the next 3 days will be “Cloudy(C) -Rainy (R) -Sunny(S)"? The initial probability of sunny is 0.6. pls 5 | List the components of Hidden Markov Model (HMM) and aie define them. 6a riven set of points. identily ¢ Jomertive clustering an r tive clustering algorithms. Di Puclidian distinee Sample No | a8 | ay | pee | eae le PO I (OR) 6b Cluster the following data points using K-means clustering algorithm with K=2 until it converges. Assume first two data points as initial cluster centroids Ci and C2. dataset = { (5,3), (10,15), (15,12), (24,10), (30,45), (85,70), (71,80), (60,78), (55,52), (80,91) } Ta Mr. X is happy someday and angry on other days. We can only observe when he smiles, frowns, laughs, or yells but not his actual emotional state. Let us start on day | in the happy state. ‘There can be only one state transition per day. It can be either to happy state or angry state. The HMM is shown below; elas 90, reo OD, Assume that qr is the state on day t and 0 is the observation on day t. Answer the following questions; Oy | dh : v)z Aun | i ae Ad ape [Or foun | | | | ay Q tree Og Hon ds te Oy bee | | | 05° frown, My thppr gam Angry gy Anes Gay duce i | Fe Ayes itn = J aap? i bt a) Hoa { | (OR) plain Baum-Welch algorithm to find the unknown parameters of a hidden Markov model (HMM). Derive the| 33 | 5 | 3 | expressions for the parameters of HMM. | Part C (X14 = 14) | i | QNo. | Questions | Marks | CO! BL | One biological application of HMMs is to determine the secondary leet structure (i.e. the general three dimensional shape) of a protein. This | ‘general shape is made up of alpha helices, beta sheets, and other [a7 | 8 | structures. In this problem, we will assume that the amino acid i composition of these regions is governed by an HMM. To keep this | 2+12 problem relatively simple, we do not use actual transition values or We will use ion probabilities below. 5 | 4 the state transition probabilities and e1 Alpha Helix | Beta Sheet Turn Other Alpha [lelix (6 OB at bu Beta Sheet 1 00 01 Turn ot a1 06 oa Other See 2 | | "Amino Acid | Alpha Helix” Beta Sheet | Turn Other i O07 ise O10 019 O15 Sn ee a) Draw this HMM as a FSM. b) Find the most likely state transition path for the amino acid sequence MLAE using Viterbi algorithm and its probability. Date: 29-09-2022 DEPARTMENT OF INFORMATION TECHNOLOGY, MIT CAMPUS 1T5012-Pattern Recognition V SEMESTE ASSESSMENT -1 Course Outcomes On completion of the course, the students will be able to: 1, Implement basic pattem classifier algorithms. Have knowledge about functionality of classifiers. Perceive the recent advancement in pattern recognition. Apply SVM and HMM algorithms for real time application Part A Implement advanced methodologies over image processing applications. (6X; Maximum Marks: 50 Have knowledge about the working principle of unsupervised algorithms. Questions: ks Write down the stages of Pattem re nition system, Given confusion matrix, define accuracy, precision, recall and speci he:perceniage of spam in the whale emails 3056. Th percentage of the Word ‘offer’ that occurs in spam e-mails 80%. The percentage of the word ‘offer’ that occurs in the desired e-mails is 10%. Find the probability of spam where the e-mail contains the word ‘offer’ using Bayes theorem. Distinguish supervised learning and unsupervised learning, Define Conditional Independence between random variables. Part B QX13 = 26) QNo. Questions Marks co | BL 6a(i) Suppose we have a two-classes problem (A, ~ A), with a single binary valued feature (x, ~ x). Assume the prior probability P(A) = 0.33. Given the distribution of the samples as shown in the following table, use Bayes Rule to compute the values of posterior probabilities of classes. Al~A x 248 | 167 ~x [82 _ [503 13,| 3 ae ation for the following, data 6 neces 5 7 — 5 1,3 i (OR) scat tomy Lage growe Yeoman [waters nt chides Toure er; Single Forest | rderaged [eum [res co | cova Yeung ig Te 3B) 43] 3 ter, [ow igh ve fake 7 [Waeged [tion Thee 7 Extended [nope [Har Te [Soule Pont [ow lew [re “| Ifthe test data is Single Parent, Young, and Low income, using Naive Bayes classifier predict whether they buy a car Derive Least Square Estimation of parameters of linear Ta 13 1,3 2 multivariate regression model. oe (oR) Derive Maximum Likelihood Estimation of parameters of] 1, ] 43] 2 7h \ | tinear regression model with Gaussian distribution. . Part C (x14=14) QNo. Questions Marks} CO | BL s 4 ]13 | 4 Ci Gulate the probability of getting chest i = Lo TaaT ae TR TOOT Roll No. [2 |p |u]e ANNA UNIVERSITY (UNIVERSITY DEPARTMENTS) B.E./ B. Tech / B. Arch (Full Time) - END SEMESTER EXAMINATIONS, NoviDec 2022 INFORMATION TECHNOLOGY 5" Semester 175012 & Pattern Recognition (Regulation 2019) Time: 3hrs CO1_| Implement basic pattern classifier algorithms. CO 2 _| Have knowledge about the working principle of unsupervised algorithms. CO3_| Have knowledge about functionality of classifiers. CO4_| Perceive the recent advancement in pattern recognition. CO5_| Apply SVM and HMM algorithms for real time applications. ‘CO6 | Implement advanced methodologies over image processing applications. BL - Bloom’s Taxonomy Levels (L1 - Remembering, L2 - Understanding, LS - Appiying, L4 - Analyzing, LS - Evaluating, L6 - Creating) PART- A (10 x 2 = 20 Marks) nswer all Questions) [No Questions Marks | CO | BL | 1 | Compare and contrast supervised and unsupervised learning 2 [23a \) uo 2__ | Adiagnostic test has 98% accuracy and 63% of all people have 2 1 | B Covid-19. If a patient tests positive, wha is the probability that the person actually has the disease? 3 | Find the Manhattan distance and Chebyshev distance between 2a (ia [es the two data points X and Y each having five attributes. , 16, 24, 16, 4] 0, 14, 9, 11,6] 4 _ | Define Grid-based clustering. Name two Grid-based clustering 2 ope ete approaches. 5 _| Distinguish feature selection and feature extraction. 2) (ase ae 6 | What is meant by dimensionality reduction? Why is it done? a) aan & 7 | Differentiate the simple Markov model and Hidden Markov model, | 2 She eae 8 | What are support vectors in SVM? What is their significance in 2 Sua prediction? ‘9 | Can a fuzzy membership be True and False at the same time? 2 [46/8 Justify your answer. 740 | Define perceptron learning rule. 2° fsaeo) oie PART- B (5 x 13 = 65 Marks} (Resitict to maximum of 2 subdivisions) @No Questions Marks | CO | BL Maximum Likelihood Estimation of parameters of linear 11) | regression model with Gaussian distribution. ae eae lee OR Income | Student | Credit Buys Rating _| Computer Tigh no fair no high no excellent | no high no fair es medium — | no fair yes low yes [fair es Tow yes | excellent [no tow yes [excellent | yes medium [no fair no 14 (b) Tow. es fair yes. 13/43 1B medium | yes | fair yes. medium | yes | excellent yes medium [no excellent | yes high yes | fair es medium | no excelent [no above dataset, if the hypothesis = "Age'''<=30, edium", "Student'yes' , "Credit Rating":fair} is given, predict whether the person buy or will not buy a computer using Naive Bayes classifier. For the given set of points, identify clusters using the simple linkage agglomerative clustering. Draw dendograms. Use Manhattan distance. Sample No X ¥. PI 1 1 Bon P2 15 15 5 a) (i 2 | P3 s a P4 3 4 rp 4 4 P6 3 35 (il) | Explain Expectation-Maximization algorithm for clustering, 5 j2 yu OR Cluster the following dala points usin; om at se 9 K-means clusterin, algorithm with K=2 for two iterations. Assume first two data points 42 (b) (i) | as initial cluster centroids C1 and C2, 8 dataset = { (5,3), (10,15), (12,12), (20,10), (30,45), (85,70), (75.80)} (ii) | Explain DBSCAN algorithm for clustering. a {2 ua Explain the Principle Component Analysis for dim 13(@) | reduction. ’ a 13 |a3 tu OR 13(b) | Explain the techniques used for feature selection. 130 [4,3 [ur Explain Baum-Welch algorithm to find the unknown parameters of 14(a) | a hidden Markov model (HMM). Derive the expressions for the| 13 |5 | 13 parameters of HMM, OR Explain the working principle of support vector machine classifier. 44 (b) | Define the hyperplanes for linear SVM. How can the margin be maximized in SVM? 13 {5 B Consider a local area network (LAN) of interconnected workstations that communicate using\ Ethernet protocols at a maximum rate of 20 Mbit/s. The two fuzzy sets given below Tepresent the loading of the LAN: o4 03,00 ae 10 10 where § represents silent and.C represents congestion. Perform union and intersection over the two fuzzy sets. 15 (a) (i) 03,05 03 08 Design perceptron network to classify the two-dimensional pattern shown in figure. The symbol ‘ * ’ indicates data to be +1 and ~‘* indicates data to be -1. The patterns are C with target +1 and A with target -1. Train the network for one epoch with initial i weights as zero, 8 as 0 and aas 1. 9 wo aan ca 46 ees “A’ Target: -1 eee “C” Target: +1 OR Find the new weights for the Back Propagation Network given below. The network is presented with the input [1, 0] and target value +1, Use a leaming rate of 0.3 and binary sigmoidal 45 (b) (i) | function for activation. The initial weights are [v11_v21 vO1] = f.5 -0.1 02], [vi2 v22 vo2]=[-01 07 04] and [wi w2 wo]= ‘[0.3 0.2 -0.4]. 13, | 46 PART- C (1 x 15 = 15 Marks (Q.No.16 is compulsory) Q. No | Questions Marks co BL | 16. (i) Consider the sequence of emotions: H,H,G,G,G,H for 6 consecutive days. Using the Viterbi algorithm find out the most likelihood of the series of hidden states.

You might also like