You are on page 1of 20
ee Ne 46MDS81 (To be filad by the candidate) (2016 Batch Onwards) COIMBATORE INSTITUTE OF TECHNOLOGY (Government Aided Autonomous institution) COIMBATORE 641 014 M.Sc. DATA SCIENCE DEGREE EXAMINATIONS, JUNE 2022 (Eighth Semester) |G6MDS81 ADVANCED PREDICTIVE ANALYTICS Time: 3 Hours Max: 75 Marks INSTRUCTIONS 4. Answer all questions in Part A and as per choice in Part B. 2. Part A and Part B questions should be answered separately in the same answer sheet. PART A (10x 2= 20) 4. Identify the qualitative feature in prediction of stock price based on company performance and economic data. 2. pate type of simpie regression model to be used for predicting the class,if raining data were generated from bivariate Gaussian distributions with uncorrelated component and different mean. 3. Write the statement of Gauss Markov theorem. 4. Which selection methods referred as greedy algorithm ? Justify the answer. 5, State the massive parallel processing system for data storage in handling anaiytical data. 6. Compare Internal analytic sandbox with external analytic sandbox. 7, White the Support Vector Machine classifier as a convex optimization problem 8. __ Relate Out of Bag sample with performance of random forest. 9. _ State the use of maximal clique in Markov graph. 40. Specify the condition for defining Markov Network as Boltzmann Machine, PARTB (x11 = 55) 41. a) 1s K-Nearest Neighbor method highly suitable for predicting the high dimensional data space ? Justify the answer, o b) Describe the procedure used for identifying the coefficients of linear model using residual sum of the square method. 4) (oR) 42. a) Generate the dataset with five data points for predicting the expenses ‘incurred in a family based on the income and family size. 4 b) Write the ‘procedure for predicting the expenses incurred in family based on the income and family size using linear least square model. ” Contd, 43. 2) Mention the impact of bias and variance trade off in model selection. b) Wite the procedure to determine the regression by successive orthogonalization for multivariate dataset. (oR) : 44, a) Generate a dataset with five data points for sanctioning the bank customer loan based on income, age, experience and loan. b) Apply forward selection procedure to generate the subset of the given features stated in Question Number 14 (a). 46, a) Describe the strategy used to generate centralized database for large data. b) Apply cloud computing to improve the efficiency of the analytic process. (OR) 16. a) Identity the appropriate strategies for handling enterprise analytic dataset along with analytic process- Justify the answer. b) Describe the method used for introducing innovation in analytic process. 17. a) Illustrate the process of determining the hyperplane for separating two classes using maximum margin technique. ) Relate the kemel functions used for classifying linear data set with kemel functions used for classifying non-linear dataset. (OR) 48. 2) Generate a random forest for classifying the weather is cool or hot based on the features outlook, temperature and humidity using entropy measures. b) Is random forest a best erisemble method ? Justify the answer. 19. a) Is the given diagram representing undirected graphical model ? Justify the answer. b) Determine the number of completed graphs and number of separated nodes existing in the above diagram. (oR) 20. a) Write a procedure to perform prediction on continuous variables using undirected gtaphical model. b) Compare Markov model with Boltzmann machine model. 4) a 4) 0) mM @) @) 6) @ 4) @) @ 6) © 6) ©) rid 16MDS81 Reg. No. (To be filed by the candidate) (201% BATCH) COIMBATORE INSTITUTE OF TECHNOLOGY (Government Aided Autonomous institution) COIMBATORE 641 014 M.Sc, DATA SCIENCE DEGREE EXAMINATIONS, MAY 2023 (Eighth Semester) AsmDs81 NCED PREDICTIVE Ss Time : 3 Hours Max: 75 Marks INSTRUCTIONS, 1, Answer all questions in Part A and as per choice in Part B. 2. Part A and Part B questions should be answered separately in the same answer sheet. PARTA (10x 2 = 20) MARKS CO BT 4. Is handwritten digit recognized using linear regression model? Justify the 2 CO1 AN answer. 2. Write the procedure to represent conditional expectation at a point of 2 CO1 U independent variables as a regression function. 3, Mention the function to predict the linear combination of parameters. 2 cot R 4. Write the dataset description for high dimensional features in financial 2 CO1 AP domain. 5. ° State the applicability of using cloud computing for analyzing the data. 2 coz u 6. _Isit feasible to have separate analytical data set? Justify the answer. 2 Coz AN 7. State the role of kemel of support vector machine for classification. 2 cos u 8. Associate number of decision trees in a random forest with number of 2 C03 U samples in a dataset. . Mention the rule for representing directed edge as an undirected edge. 2 cos 10. Specify the condition for referring clique as maximal clique. 2 co3 PARTB_ (8x11 =55) 14 a) r model applicable for classifying the 7 | CO1 AN for a loan in a bank as low risk customer or high risk customer. b) Justify the answer given for the question stated in Question 4 CO1 AN No.11(@) (OR) 12 a) Why K-Nearest Neighborhood model not applicable for 4 CO1 U handling high dimensional featured dataset. b) Describe the different functions to be used as the input 7 CO1 U features for making linear models for handling high dimensional input features. 13 a) State and explain Gauss-Markov theorem, 5 cor U b) ‘Apply the orthogonalization procedure for predicting the price 6 CO1 AP of the house using various features such as building area(in square feet), number of floors, number of rooms, wood work area, type of wood, electricity connections available, bore well (if any), compound wall(f applicable) and height and width of compound wall. (OR) Contd... | 14 15 16 17 18 19 20 b) a) b) a) b) ) b) a) b) a) b) Compare the applicability of lasso with ridge regression models, in the context of reducing the error during the analyzes of the data. Apply principal component regression model for predicting the number of independent variables required for analyzing the customer transactions in retails store fo predict the demand of the product. Compare centralized with distributed data for performing the analysis. Write the consequence of applying external analytic sandbox for predicting the classification of weather as suitable for playing tennis or not based on various features. (OR) Write the tasks involved to generate the analytic dataset for prediction of classification or regression. Determine the innovative predictive strategy for analyzing the data in the production of a machine in manufacturing demain. ‘Write the procedure to classify support vector machine as a maximum classifier, Is radial basis kemel applicable for the analyzing the text analysis problem? Justify the answer. (OR) How out of bag sample are used to analysis the performance of the random forest. Write the procedure to generate the random forest for classifying the given retail store data as regular customer or irregular customer. ‘Apply Markov graphical model to determine the optimal path for the travelling salesman from source to destination place. Write the modified regression procedure to generate the Gaussian graphical model with known structure. (OR) Compare Gaussian and Boltzmann machine model for analyzing the data. Describe the applicability of hidden nodes in Boltzmann structure for describing the data. cot coz coz cos co3 co3 cos cos AP AP AN AN AP AP Rona, . 16MDS82 (To be filed by the candidate) (2016 Batch Onwards) COIMBATORE INSTITUTE OF TECHNOLOGY (Goverment Aided Autonomous Institution) COIMBATORE 641 014 M.Sc. DATA SCIENCE DEGREE EXAMINATIONS, JUNE 2022 (Eighth Semester) 46MDS82 DATA PRIVACY AND SECURITY Time : 3 Hours Max: 75 Marks INSTRUCTIONS 1. Answer all questions in Part A and as per choice in Part B. 2. Part A and Part B questions should be answered separately in the same answer sheet. PARTA (10x 2=20) 1, What is meant by CIA traid ? 2. Why Avalanche effect is important in cryptographic algorithms 3. Find GCD of 1176 and 1492 using Euclidean algorithm. 4, What is meant by primitive root 7. Give’an example. 5. Determine Euler's Totient Function of 213. 6. For the point P (11, 2) defined in the curve E13 (1, 1), Whatis -P ? 7. - Why hash function is called as one-way function ? 8, Whatis malicious code ands how it causes harm to resources 7 9. What triggers a virus to start replicating ? 10. Define integrity. PART (6x11 = 55) 11. a) What are Passive and Active Attacks ? @) b) Bob wants to send a message to John. Both Bob and John uses secret key as 3. Now Bob sends a message “RSHQLQVWDJUDP" to John and John recovers the original message and reads it as [Let A=0, ....,Z=25) Oo ) Encrypt the message “meet captain tonight” using Playfair cipher with the key autonomy’. © (OR) 42. a) How cryptographic systems are characterized along three independent dimensions ? @) by iiustrate with a neat diagram how Feistel structure is used in DES algorithm to encrypt and decrypt messages. ® 43. a) Find GOD and inversie of 240, 46 using Extended Euclidean Algorithm. ® b) Solve using Fermets theorem : 4° mod 41. @ ¢) Solve using Euler's theorem : 3° mod 13. : @ (OR) Contd... 14. 15. 18. 47. 18. 19. 20. a) Explain in detail the various steps involved in Advanced Encryption Standard. b) Given the values of q= 23, a=7, Xa=3, and Xs = 6, show how entity A and B securely exchange the secret key using Diffie-Hellmen key exchange Protocol. a) Ina RSA public key system, intruders intercepts the cipher text C = 26 send to the user whose public key is e = 7, n= 33. What is the plain text M 7 b) Analyze the advantages of ECC over RSA algorithm. (OR) a) Describe how Message Authentication Code authenticates messages being sent to users, b) Illustrate how Secure Hash Algorithm generates hash code with neat diagram. a) Show how PGP provides confidentiality and authentication service in electronic mail and file storage applications. b) What is IPSec ? Mention its benefits. (OR) a) Describe briefly any two targeted malicious code. b) Compare transient and resident virus. a) Explain two-phase update in databases with an example. b) Discuss the approaches to multilevel security of databases. (OR) a) Identify how data analytics can be used to preserve privacy. b) Write the steps to protect against privacy loss. ©) 6) a (4) 4) ™ ©) 6 a a © 6) © 6) 10. 41. 12. Reena! 416MDS82 (To be filled by the candidate) (2019 Batch) COIMBATORE INSTITUTE OF TECHNOLOGY (Government Aided Autonomous institution) COIMBATORE 641 014 M.Sc, DATA SCIENCE DEGREE EXAMINATIONS, MAY 2023 (Eighth Semester) 6MDS82 DATA PRIVACY AND SECURITY Time : 3 Hours Max: 75 Marks. INSTRUCTIONS 1. Answer all questions in Part A and as per choice in Part B. 2. Part and Part B questions should be answered separately in the same answer sheet, PARTA (10x 2=20) war en co Differentiate active and passive security attacks. 2 Ui COz How cryptography algorithms are classified? 2 U_ co2 Are the numbers 24 and 225 relatively prime numbers? Justify your ansiver. 2 AN CO1 Solve using Euler's theorem: 7 mod 10. 2 AP CO1 For the point P (12,5) defined in the curve E13(1, 1), What is -P? 2 U_ cos Name a one-way function. Why it is called so? 2 wu cas List any four types of inadvertent flaws. 2 R cca Differentiate Worm and Virus. atau Od) White any four advantages of databases over file system. 2 R cos ‘What is meant by element integrity? 2 R cos PARTB (6x 11=55) a) Describe the six security services defined by OSI standard. 6 uU_ coz b) Write a brief note on CIA traid. 5 R coz (OR) a) Encrypt the message “meet b” using 2x2 Hill cipher with the key 6 SY CO3 By ef a b) Encrypt the message “meet me after the party" using Playfair cipher 5 SY COS with the key “predominant. Contd... 13, 14, 15. 16. 47. 18. 19. 20. a) b) a) b) a) b) a) b) a) ») a) b) b) a) b) Compare the model of Symmetric and Asymmetric encryptions. Explain in detail the various steps involved in Advanced Encryption Standard Algorithm. (OR) Find GCD and Multiplicative inverse of 635, 234 using Extended Euclidean Algorithm, Ina RSA public key system, intruders intercepts the cipher text o=37 ‘sent to the user whose public key is e=3, n=55. Find the plain text M Given the elliptic curve y?=x°+x+1 and p = 23, determine any five points on the curve. Explain how in an Elliptic Curve Cryptosystem, two users A and B selects their private key and computes their public key. (OR) List any four applications of hash algorithms. Illustrate how message digest is generated in Secure Hash Algorithm with neat diagram. Describe how confidentiality and authentication services for electronic mail is provided by Pretty Good Privacy with a neat diagram. Compare transport and tunnel mode in IPSecurity. (OR) Explain how viruses attach and gain control over a system? Discuss about any three targeted malicious code. Write short notes on Authentication and Privacy. Write the Steps to Protect Against Privacy Loss. (OR) Explain how machine learning techniques are used to provide security? Discuss in detail how security techniques can be used in Machine Learning scenarios to preserve privacy? AN sy sy sy AN cos coz cot cot, co3 cot, cO3 cos, co3 cos co4 co4 co4 co4 cos cos ala : 16MDS83 Reg.No (To be filled by the candidate) (2016 Batch Onwards) COIMBATORE INSTITUTE OF TECHNOLOGY (Government Aided Autonomous Institution) COIMBATORE 641 014 M.Sc. DATA SCIENCE DEGREE EXAMINATIONS, JUNE 2022 (Eighth Semester) DATA VI ‘TION Time : 3 Hours Max: 75 Marks INSTRUCTIONS 4. Answer all questions in Part A and as per choice in Part B. 2. Part A and Part B questions should be answered separately in the same answer sheet. PARTA (10x 2=20) 1. Depict the interactions between the seven stages of visualization. 2. "Do you think “knowing your audience" makes significant difference in visualization 7 List out the three classes used in tree map library. State triple (0, C, F) representation of continuous data. Mention the difference between global and local illumination. ‘What is color banding effect and why it is nat desirable ? Define fiber clustering 7 2 NOM a w illustrate the importance of visualizing quality metrics ? 9. Write a code to Append a new DOM element. 10. Define the terms legend and axes in D3 charts. PARTB (5x 11=55) 11. a) Explain how to use text as input in a complicated graph ? 6) ) Explain the terms Conriections and Correlations. How scatterplots are used for representation. Give example. ® (OR) 12. a) State the importance of Data Visualization. @) b) Explain in detail about the seven stages of visualization with an example. ®) 18. a) Define Visualization Pipeline. @) b) Discuss the Discrete data representation in visualization applications with a neat diagram. @ (OR) 14, _ Describe about various stream objects available for visualizing the trajectory of input object in a vector field over a given time interval. (4) Contd... 15. __iflustrate the methods in reducing the computational complexity of contouring, (1) (OR) 46. With an exampie explain the following terms a) Table Visualization @) b) Visualization of Relation @ i c) Multivariate Data Visualization @) ! 4) Text Visualization ® 47. Explain the difference ‘between Infovis and Scivis using a real-world example. Explain with respect to dataset, data domain and data attributes. ) 1 (OR) | | 18. — Write a code to draw blue circle in SVG with 60 radius using D3 with Data Object. (a1) 19, Define Data Binding. Explain the commands select, select All, append, exit in D3 js with example. ay i (OR) 20. a) Explain Data — Joins in D3 with example. (6) b) Write a code to create following list element using Data Object in D3 * MSc Data Science M.Sc Software Systems M.Sc Decision and Computing Science M.Sc Artificial Intelligence and Machine Learning MCA (6) cared 16MDS83 (To be filled by the candidate) (2019 Batch) COIMBATORE INSTITUTE OF TECHNOLOGY (Government Aided Autonomous Institution) COIMBATORE 641 014 M.Sc, DATA SCIENCE DEGREE EXAMINATIONS, MAY 2023 (Eighth Semester) DATA VISUAL Time :3 Hours Max: 75 Marks INSTRUCTIONS 4. Answer all questions in Part A and as per choice in Part B. 2. Part A and Part B questions should be answered ‘separately in the same answer sheet. PARTA (10x 2=20) MARKS BTL CO 1. Draw the interactions between the seven stages of data visualization. ot io col 2. “Avoid the All-You-Can-Eat Buffet’ — Justi the response with an example. 2 10 cor 3. Whatis a tree map? 2 Lo co2 4. Mention the difference between local and global illumination. 2 HO CcO3 5. What do you mean by the term texture mapping? 2 Lo cot 6. Whatis color banding effect and why is it not desirable? 2 0 coe 7. List the goals to design an effective color map for scalar data. See eal 8. Give the importance of fiber clustering. Hei cay 9. Define the terms legend and axis in D3 charts. 2 Lo cot 10. — Give a brief mention of the Clamp() used in D3 linear scale 2 10 Cor PARTS (6x11 =55) 11. a) _List out the seven stages of visualization. seetiateat b) Using an example explain the seven stages of visualization. ee epe (OR) 12 4) How scatterplot maps are used for data representation? 6 10 Cot b) Describe the terms connections and correlations to represent the 6 lo coi data 13. 14, 15. 16. 17. 18. 19. 20. a) b) a) b) a) b) a b) a) ») b) How will you recursively print the tree content to represent the data? Give example. How will you perform parsing the data in a HTML file? {OR) Differentiate between Infovis and Scivis with respect to data set . How will you visualize vector PCA Information? ‘Sketch and outline the functional view of the visualization pipeline. How will you perform the data filtering and enrichment in the visualization process? (OR) Explain the mapping process done in visualization pipeline with a Suitable example. Describe the process of reducing the computational complexity of contouring with example. (OR) Explain about table visualization with an example. How will you perform content based visualization in a text document? Write a code to draw a green square whose side is 30 using D3 with data object. How wilt you label the data points with text elements in SVG? (OR) How will you create and incorporate dynamic scales in D3? Define Data Binding, Expiain the commands select all, append and text in D3 with example. 1 "1 HO. HO. 10 Lo Io HO 10 HO. HO. HO co2 co2 cot co2 cot co2 coz co2 cot " 12, 13. Beene 46MDSE7 Reg. No. (Tobe filled by the candidate) (2019 Batch) COIMBATORE INSTITUTE OF TECHNOLOGY (Government Aided Autonomous institution) COIMBATORE 641 014 M.Sc. DATA SCIENCE DEGREE EXAMINATIONS, MAY 2023 (Eighth Semester) 46MDSE7 IMAGE PROCESSING Time ; 3 Hours Max: 75 Marks INSTRUCTIONS. 4. Answer all questions in Part A and as per choice in Part B. 2. Part A and Part B questions should be answered ‘separately in the same answer sheet, PARTA (10x 2 = 20) MAR’ CO” RET K 1. Uist out any four operations performed in digital image processing. 2 cor R 2 Define Image. 2 corR 3. What is meant by Averaging Filter? 2 coz R 4. Why Isotropic Filters are known as Rotation invariant Filter? 2 coz R 5. Recall the three methods of segmentation. 2 COB U 6. How Periodic functions are expressed? 2 cox u 7. State the need for Image Compression Technique. 2 co4 u 8 Name the 2 processes used to convert the continuous sensed data into 2 CO4 R digital form. 9. List any 2 applications of Motion Tracking. 2 cos U 10. Define Classification System. 2 COB R PARTB (5x 11 = 55) u a) _ Elaborate the role of Sensor Arrays in the process of image Acquisition § COt ») _. Describe the various types of connectivity between pixels with suitable 5 CO1 U example (OR) @) With neat diagram explain CMY and CMYK Color Models in Digital 5 Col U Image Processing b) Let p and qbe the pixels at co-ordinates (10,12) and (15,20) 6 cot AP respectively. Find out which distance measure gives the minimum distance between the pixels a) Perform contract stretching and Dee eta 7 coz AP given below v tie a i gen boow 4, (Given 1123, 7925 & S128, $226) 3 6 4 6 2 2 6 5 ed 6 4 q b) Consider the following image, find what will be the new values of 4 CO2 AP the pixel (2,2) if smoothing is done using 3 * 3 neighbourhood by applying the following techniques }) Mean Filter ii) Median Filter iii) Max Filter iv) Min Filter (OR) 14. 15. 16. 47. 18. 19. 20. a) b) a) b) a) b) a) b) a) ») Perform the histogram equalization of the given image Fe4 acacia 4 34 5 43 395 5 5 3 Si aS aiS: Describe in detail about region splitting and merging technique in image processing. ‘Summarize the steps involved in Canny Edge Detection Algorithm. (OR) For the following input image perform multi-resolution analysis by applying harr wavelet transform 100 50 60 150 20 60 40. 30, 50 90. 70. 82 74. 66. 20. 58 Apply region growing on the following image by assuming the threshold vaiue b <=4, 6 In a Communication System, the source transmits five different messages, say x1,x2,x3,x4 and x6 with the probabilities of 0,3, 0.15, 0.25, 0.05, 0.25 respectively. If the number of symbols used to code these messages is 2, then find the code word for each message and the coding efficiency by applying, Huffman Coding Technique. (OR) With neat diagram, explain the Lossless Predictive coding process. Find the opening and Closing of the given grey scale image by considering the Structuring element as B = [1 1 1] X= [10000 01000 00100 00010 ooocty ‘Sketch a diagram for the steps involved in classification of an image. What are the advantages of representing regions in the form of graphical representation. (OR) Explain how chain codes and skeletons are used for representing shapes. Define the eccentricity of the boundary, " co2 cos cos cos co4 cos cos AP AP AP AP AP Satie 16MDSE7 Reg. No. (To be filled by the candidate) (2018 — 2020 Batch) COIMBATORE INSTITUTE OF TECHNOLOGY (Government Aided Autonomous Institution) COIMBATORE 641 014 M.Sc. DECISION AND COMPUTING SCIENCES DEGREE EXAMINATIONS, NOVEMBER 2022 (Ninth Semester) GMDSE7 IMAGE PROCESSING Time : 3 Hours Max: 75 Marks INSTRUCTIONS, Answer all questions in Part A and as per choice in Part B. 2. Part A and Part B questions should be answered separately in the same answer sheet (10x 2=20) Marks Bloom's Course ‘Texonemy Outcome 1, White the representation of a digital image and define it @ R cor 2, List out the tasks performéd by image interpolation. @ R cor 3. Define the purpose of image registration. @ vu co 4, Sketch the general approach of image transforms, Q R coz 5. Compare spatial and frequency domain firing @ vu coz 6. Give the difference between lossless and lossy compression @ vu cos 7. raw the histogram for bright and low contrast image. @ uv coz 8. Whatiis point detection in image processing ? @ Rk cos 9. Define region-based segmeniation. (2) R cos 40, State the principle of feature extraction in image processing @ v cos PARTE (6x11 =85) 411. a) Describe image formation in the eye, brightness adaptation and discrimination in brief. ® vu co b) Classify the three types image sensors used for image acquisition © u co (OR) 42, a) Identify the following properties on the pixel eloments in a digital image. (i) Adjacency (2) (i) Connectivity @ u_ cor (i) Region @) (iv) Boundary @) b) Narrate image negative transformation function @ R~ co2 + 13. a) Analyse the basic mathematical tools used in digital image processing. @ aN cor ») Formulate the types of mean fiters used in spatial fitering : (cen cae (OR) Contd 14, 45. 16. 7, 18. 19. 20. Consider the following intensity distribution of a 3-bit image (L = 8) of size 64x64, where the intensity levels are integers in the range of [0, L- 1] =[0, 7] Perform histogram equalization and infer the results of original and equalized histogram. a) Summarize the procedure for discrete histogram specification. b) What is kernel ? Demonstrate 1-D spatial correlation and convolution result for the kernel w= [7 3 135] (OR) a) State the role of thresholding in applications of image segmentation and explain the properties of intensity thresholding. b) Examine the role of noise, illumination and reflectance in image thresholding with its corresponding histogram. a) Explain how lines are detected in an image using the Hough transform ? b) Discuss how edges are linked through Hough transform 7 (OR) a) Classify the various edge models in image segmentation b) Compare first and second order derivatives for smoothes an image using example. 2) What is image compression ? Explain JPEG image compression standards. b) Discuss about multiresolution theory for signal processing using the functions involved in wavelet transforms. (OR) Narrate short notes on the following topics. a) Video Motion Analysis b) Image Fusion ) Steganography (it) (6) © 6) © (5) (5) (6) 4) @) @) 6) ap ap AN AN co2 coz coz 05 cos co4 co4 cos co3 cos a A6MDSE7 Reg. No. (To be filled by the candidate) (2016 Batch Onwards) COIMBATORE INSTITUTE OF TECHNOLOGY (Government Aided Autonomous Institution) ‘COIMBATORE 641 014 M.Sc. DATA SCIENCE DEGREE EXAMINATIONS, JUNE 2022 (Eighth Semester) 16MDSE7 IMAGE PROCESSING Time : 3 Hours Max: 75 Marks 10. 11. 42. 13. INSTRUCTIONS 4. Answer all questions in Part A and as per choice in Part B. 2. Part A and Part B questions should be answered separately in the same answer sheet. PARTA (40x 2 = 20) Define Image processing. List out the typical operations of digital image processing. Which one is best known filter among Order-Statistic (Nonlinear) Filters ? List out the three techniques of Piecewise-Linear Transformation. Most of the segmentation algorithms are based on the following two basic properties of intensity values namely & List out the types of Data Redundancy State the components of general compression system model. Differentiate Lossless and Lossy compression. Define Compression ratio. “Image recognition is also known as computer vision’ - TRUE or FALSE ? PARTS (6x 1165) a) Describe the key stages in digital image processing with a neat diagram. ©) ) State the two processes involved in creating a digital image, by converting the continuous sensed data into digital form @ (OR) a) Elaborate the role of sensor strips in the process of image acquisition ® b) Difference high dynamic range and low dynamic range. ) Perform histogram equalization for the following image segment. (ty Oar aeis eed fade 2 tee OEE Foye 6 3 2 4 1 aera ied (OR) Contd. 14. @) Illustrate the difference between first and second derivatives of a digital function of a horizontal intensity from an image. ') Compute the median value of the pixel of the image A (x, y) shown below. AG y= 10 30 5 20 200 20 15 10 30 15, a) Discuss the various approaches for detection of edge in an image. ») Define global thresholding. (OR) 16. Describe in detail about region spiiting and merging image in segmentation technique. 17. Calculate the binary Hufiman code for one following symbol whose probabilly of ‘occurrence is given below and also find the expected code length for this encoding. ‘Symboi Probability a 0.49 a 0.26 a 0.11 as 0.03 a5 0.03 as 0.02 a; 0.01 (oR) 18. a) Differentiate between encoder and decoder with necessary block diagrams. b) State the uses of wavelets. 19. a) Explain how chain codes and skeletons are used for represent ting shapes. b) define shape number. (OR) 20. a) Discuss about image classification and its types. 5) Illustrate the types of color models with examples. @ 6) 8) 3) (1) (11) (8) (3) @ @) © ©) Pena oo 10. 1. 12, GA2023 Reg. (To |. No. be filled by the candidate) COIMBATORE INSTITUTE OF TECHNOLOGY (Government Aided Autonomous Institution) COIMBATORE 641 014 1 A6MDSE10 (2019 Batch) M.Sc. DATA SCIENCE DEGREE EXAMINATIONS, MAY 2023 (Eighth Semester) AGMDSE10 NATURAL LANGUAGE PROCESSING Time : 3 Hours a) b) a) b) INSTRUCTIONS Max: 75 Marks: 4. Answer all questions in Part A and as per choice in Part B. 2. Part A and Part B questions should be answered separately in the same answer sheet. PARTA Define ambiguity. Differentiate words with transducers. ‘Mention the use of tag sets. Discuss on error analysis. What do you mean by text normalization. Define phonetics. Mustrate on finite state machines. Write a note on CKY parsing. Differentiate semantics with pragmatics, List any two machine translation techniques. PARTB Explain speech and language processing in detail Describe about the challenges in NLP_ (OR) illustrate on the steps to be followed for detecting and correcting spelling errors: ‘Write a note on tokenization NVNNNE N R COs (5x 11 = 55). 6 R cot 5 U_ cot Contd... 13. 14. 15. 16. 17. 18. 19. 20. a) b) a) b) a) b) 5) ») Define POS tagging and expiain its types in detail (OR) Explain statistical tanguage model and find the probability of the test sentence ‘P ( they play in the big garden) in the following training set using bi-gram model There is a big garden Children play in the garden They play inside beautiful garden List the problems associated with the bi- gram model. Differentiate between phonetic analysis and prosodic analysis. (OR) Write short notes on feature extraction. Mention the steps to be followed for HMM to speech Discuss about grammar rules and tree banks. Explain about dependency grammar in detail. (OR) Enumerate on dynamic programming parsing methods. ‘Compare and contrast Computational semantics with Lexical semantics in detail. (OR) Mention the techniques to improve retrieval performance. Explain about the benefits of machine translation. " 1" “a "1 uu coz AP COZ ue co2z R cos R cos u_ -CO3 u_-cos uU_-CO3 AP CO4 R co4 u COs uUCO5

You might also like