answer all questions as instructed below.

attend all questions.4 Computer Vision (a) Explain why such a tiny number of 2D Gabor wavelets as shown in this sequence are so efficient at representing faces, and why such wavelet-based encodings deliver good performance in "appearance-based" algorithms for face recognition. What role do such encodings have in representing different facial expressions? What sort of neural evidence is there for such encodings in human vision? Number of Wavelets 16 52 116 216 original [6 marks] (b) Explain the "receptive field" concept as used both in CNNs (convolutional neural networks) and in visual neuroscience, and explain the role of trainable connections. Why is the concept of convolution relevant? Roughly how many layers are in the 'FaceNet' CNN, and how many connection parameters must be trained? By comparison, in the visual cortex of the brain, typically how many synapses are there per neurone, and what is the total length of "wiring" (neurite connections) per cubic-millimeter? Can connection updates be a basis for computer vision? [6 marks] (c) In relation to the image formation diagram shown below, explain: (i) the concept of a reflectance map; (ii) what is a specular surface; (iii) what is a Lambertian surface; and (iv) what is surface albedo. Give the defining relationships for the amount of light from a point source that is scattered in different directions by such illuminated surfaces, and describe the inferences that a vision system must make with them. [8 marks] .->q'l :r )-- ?rxi The definitions of the angles i, , and g 5 (TURN OVER) CST2.2019.8.6 5 Cryptography (a) The Tripos Encryption Standard (TES) is a block cipher optimized for use on UGPs ("undergraduate processors"). It operates on 4-bit blocks, written as hexadecimal digits (e.g., a 9 = 3). For one particular key K, it implements the following permutation: m 0 1 2 3 4 5 6 7 8 9 a b c d e f EK(m) 1 b 5 c 7 e 2 a 4 9 f d 0 3 6 8 Using this key K, decrypt the following three ciphertexts according to the indicated modes of operation. [Note: the XOR table at the bottom of this page may be of use.] (i) ECB: 188b06 [2 marks] (ii) CBC: 301b2 [3 marks] (iii) CFB: 10f6d [3 marks] (b) State four advantages that counter mode has over either CBC or CFB mode. [4 marks] (c) Using the same K as in Part (a): (i) Show that the CBC-MAC tag for message 1234 is d. [3 marks] (ii) Demonstrate that CBC-MAC with a given K is not collision resistant, by showing how to find another message, of the form 1x04, that results in the same CBC-MAC message tag (without iterating over different candidates for 4-bit block x). [5 marks] UGP XOR accelerator: 0 1 2 3 4 5 6 7 8 9 a b c d e f 0 0 1 2 3 4 5 6 7 8 9 a b c d e f 1 1 0 3 2 5 4 7 6 9 8 b a d c f e 2 2 3 0 1 6 7 4 5 a b 8 9 e f c d 3 3 2 1 0 7 6 5 4 b a 9 8 f e d c 4 4 5 6 7 0 1 2 3 c d e f 8 9 a b 5 5 4 7 6 1 0 3 2 d c f e 9 8 b a 6 6 7 4 5 2 3 0 1 e f c d a b 8 9 7 7 6 5 4 3 2 1 0 f e d c b a 9 8 8 8 9 a b c d e f 0 1 2 3 4 5 6 7 9 9 8 b a d c f e 1 0 3 2 5 4 7 6 a a b 8 9 e f c d 2 3 0 1 6 7 4 5 b b a 9 8 f e d c 3 2 1 0 7 6 5 4 c c d e f 8 9 a b 4 5 6 7 0 1 2 3 d d c f e 9 8 b a 5 4 7 6 1 0 3 2 e e f c d a b 8 9 6 7 4 5 2 3 0 1 f f e d c b a 9 8 7 6 5 4 3 2 1 0 6 CST2.2019.8.7 6 E-Commerce (a) Describe five ecommerce business models. [5 marks] (b) Describe five things to consider when internationalising an ecommerce business. [5 marks] (c) Does the nature of the Internet naturally lead to dominant firms in ecommerce markets? [10 marks] 7 (TURN OVER) CST2.2019.8.8 7 Hoare Logic and Model Checking Consider a programming language that consists of commands C composed from assignments X := := alloc(E1, . . . , En), heap assignment [E1] := E2, heap dereference X := [E], disposal of heap locations dispose(E), the no-op skip, sequencing C1; C2, conditionals if B then C1 else C2 (where B is a boolean expression), and loops while B do C. null is 0 (a) Explain informally what it means for a separation logic partial correctness triple {P} C {Q} to be valid. [3 marks] (b) Explain informally what it means in terms of the executions of C for the separation logic partial correctness triple {>} C {} to be valid. [2 marks] (c) Recall the list representation predicate list: list(t, []) = (t = null) list(t, h :: ) = y.((t 7 h) ((t + 1) 7 y) list(y, )) We write [] for the empty mathematical list; h :: for the mathematical list the head of which is h, and the tail of which is ; ++ for the concatenation of mathematical lists and ; [i] for the i-th element of the list , starting at 0; and [k, . . . , n] for the ascending list of integers from k to n, including k and n. Give a proof outline, including a loop invariant, for the following triple: {N = n N 0} X := null; while N > 0 do (X := alloc(N, X); N := N 1) {list(X, [1, . . . , n])} [4 marks] (d) Also recall the partial list representation predicate plist: plist(t, [], u) = (t = u) plist(t, h :: , u) = y.((t 7 h) ((t + 1) 7 y) plist(y, , u)) Give a loop invariant for the following list sum triple: {list(X, )} Y := X; N := 0; while Y 6= null do (M := [Y ]; N := N + M; Y := [Y + 1]) n list(X, ) N = Plength()1 i=0 [i] o [4 marks] (e) Give a loop invariant for the following list concatenation triple: {list(X, ) list(Y, )} if X = null then Z := Y else Z := X; U := Z; V := [Z + 1]; while V 6= null do (U := V ; V := [V + 1]); [U + 1] := Y {list(Z, ++ )} [5 marks] (f ) Describe precisely a stack and a heap that satisfy list(X, [1, . . . , 3]). [2 marks] 8 CST2.2019.8.9 8 Information Theory (a) Consider the four-state Markov process in the graph below. It emits the eight letters {A, B, C, D, e, f, g, h} with probabilities and changes of state as shown, but note the sequence constraints. (For example, an A can only be followed by a B or an e.) Letter emissions with clockwise state transitions occur with probability , and the others with probability 1 , where 0 < < 1. p = p = p = p = 1 p = p = 1 p = 1 p = 1 "B" "C" "A" "D" "h" "g" "f" "e" (i) First imagine a one-state Markov process that emits any of eight letters with equal probabilities. What is its entropy? [2 marks] (ii) For the four-state Markov process shown with parameter , what is the long-term probability distribution across the eight letters? [4 marks] (iii) In terms of parameter , what is the overall entropy H() of this four-state Markov process? [2 marks] (iv) Sketch a plot of H() as a function of . Compare its maximum value with your earlier answer in Part (a)(i) for a one-state Markov process that also emits eight letters, and explain the difference, if any. [4 marks] (b) Is it possible to construct an instantaneous code (a code possessing the prefix property) for a five-letter symbol set using codewords whose lengths in bits are: 1, 2, 3, 3, and 4 bits? Justify your answer by stating the relevant condition. [4 marks] (c) Provide an operation in linear algebra that involves simply the multiplication of a matrix by a vector, which describes the Discrete Fourier Transform of a discrete sequence of data f[n] = (f[1], ..., f[N]) to obtain Fourier coefficients F[k] = (F[1], ..., F[N]). Define the elements of the (N N) matrix and give the computational cost of the operation in this vector-matrix form. [4 marks] 9 (TURN OVER) CST2.2019.8.10 9 Machine Learning and Bayesian Inference In designing a system to perform linear regression with noisy data, you feel that the noise is not modelled well by the usual normal density, and wish instead to use the Cauchy density p(x) = 1 2 (x ) 2 + 2 having parameters and . (a) Denote the weights of your model by the vector w. Given a set s of m examples, each consisting of a d-dimensional vector x and corresponding label y, find an expression for the likelihood p(y|w) where y T = (y1, . . . , ym). State any assumptions you make. [6 marks] (b) In addition to the unusual noise density, you have some knowledge of the problem at hand suggesting that some of the parameters in w are likely to be close to known values, whereas the others have no such constraint. Suggest a suitable prior density p(w) that could be used to model this. You should assume that the weights are independent for the purposes of designing a prior, and you may use the fact that the normal density is p(x; , 2 ) = 1 22 exp 1 2 2 (x ) 2 . [5 marks] (c) Using your answers to parts (a) and (b) derive a maximum a posteriori (MAP) learning algorithm for the problem. Should your algorithm require derivatives you may state them without working them out in full. [6 marks] (d) Suggest a way in which any parameters other than w might be either given an effective value or removed from consideration. Give a single advantage and a single disadvantage of the method you suggest. [3 marks] 10 CST2.2019.8.11 10 Mobile and Sensor Systems A team of geologists want to set up a system to monitor an area close to a dormant volcano for seismic movements. They have a single basestation which will connect to cellular infrastructure and transmit the data back to their server. They plan to scatter sensor nodes in the area to be monitored. (a) The team has decided to use SMAC for their Medium Access Control (MAC) layer routing protocol. Explain the advantages and limitations of using SMAC in such a system. [5 marks] (b) Illustrate a multihop routing solution. Describe in detail the protocol you would use for this system indicating advantages and limitations. [4 marks] (c) Describe a single hop solution using an IoT protocol and explain its strengths and weaknesses (include details of the protocol). Discuss whether this solution is better or worse than the solution in Part (b). [5 marks] (d) Now assume that the sensor nodes are mobile (e.g., deployed on autonomous mobile ground platforms). (i) Explain the benefits of using autonomous mobile sensor nodes in this application. [2 marks] (ii) Describe a mechanism by which the mobile sensor nodes can collaborate to improve the monitoring information obtained. Include details of your estimation framework, and explain how it would be used to inform the motion planning of your mobile sensor nodes. [4 marks] 11 (TURN OVER) CST2.2019.8.12 11 Optimising Compilers The following C-style code from an untyped language is analysed by a compiler, where the work() function is assumed to have no side effects. 1 c = &b; 2 *c = &c; 3 a = c; 4 c = &d; 5 if (v == 0) 6 *c = **a; 7 else 8 *c = *b; 9 *a = &a; 10 work(a); 11 work(c); (a) Describe alias analysis and the transformations it enables. [4 marks] (b) Summarise Andersen's analysis and calculate the points-to set, pt(x), for each pointer, x, within the C-style code above. [9 marks] (c) Describe the reason that the analysis overestimates some of the sets in the answer to Part (b). [2 marks] (d) Now assume that the work() function may alter memory locations reachable through its argument. Explain why the two calls to work() in lines 10 and 11 cannot be executed concurrently using the analysis from Part (b), but can be based on the answer to Part (c). [5 marks] 12 CST2.2019.8.13 12 Principles of Communications (a) The Border Gateway Protocol (BGP) uses attributes to enforce transit relationships for Outbound route filtering, and to enforce the order of route preference between customers, peers and providers. Key attributes, in precedence order, are as follows (from highest to lowest): Highest Local Preference Shortest ASPATH Lowest MED i-BGP < e-BGP Lowest IGP cost to BGP egress Lowest router ID Discuss the basic use of the relevant BGP mechanisms and their use of attributes for backup and for traffic engineering. (i) Explain the different attributes. (ii) Explain the mechanisms that are used to determine paths within an AS. (iii) Explain the mechanisms that are used to determine paths between ASs. [5 marks each] (b) BGP announces and withdraws prefixes between Autonomous Systems so that different domains can route traffic amongst themselves. The dynamics of BGP advertisements can be impacted by intra-domain routing, so that if a route flaps, for example because of an intermittent fault on a link or router, this can be exported to the whole Internet. Simple techniques of fixing an interval for advertisements, and punishing routers that exceed that rate have been proposed. Describe very briefly how one might use a control theoretic approach to provide stable damping of this effect dynamically, and possibly, more efficiently. [5 marks] 13 (TURN OVER) CST2.2019.8.14 13 Types Recall the three judgements for classical propositional logic: (a) ; ` e : A true - e is a proof of type A (b) ; ` k : A false - k is a refutation of type A (c) ; ` he |A ki contr - he |A ki is a contradiction at type A Here, contains the true assumptions, and are the false assumptions. In this question, we will extend classical propositional logic with support for the implication or function type operator A B. (a) Give a proof term and inference rule for a proof of type A B. [4 marks] (b) Give a proof term and inference rule for a refutation of type A B. [Hint: how is implication encoded in classical logic?] [4 marks] (c) Give a reduction rule for contradiction configurations of the form he |AB ki. [4 marks] (d) (i) State the preservation theorem for classical logic. [2 marks] (ii) Give the proof of preservation for the case of the new rule defined above. You may assume that weakening, exchange and substitution all hold. [6 marks]6 Computation Theory (a) (i) Give an inductive definition of the relation M = N of -conversion between -terms M and N. [3 marks] (ii) What is meant by a term in -normal form? [1 mark] (iii) If M and N are in -normal form, explain why M = N implies that M and N are -equivalent -terms. [2 marks] (You need not define notons such as -equivalence and capture-avoiding substitution.) (b) Show that there are -terms True, False and If satisfying If True M N = M and If False M N = N for all -terms M and N and with True 6= False. [4 marks] (c) Define Curry's fixed point combinator Y and prove its fixed point property. [3 marks] (d) Consider the following two properties of a -term M: (I) there exist -terms A and B with M A = True and M B = False (II) for all -terms N, either M N = True or M N = False. Prove that M cannot have both properties (I) and (II). [Hint: if M has property (I), consider M (Y (x. If (M x) B A)).] [4 marks] (e) Deduce that there is no -term E such that for all -terms M and N E M N = ( True if M = N False otherwise [3 marks] 7 (TURN OVER) CST1.2019.6.8 7 Foundations of Data Science (a) Let X1, . . . , Xn be independent binary random variables, P(Xi = 1) = , P(Xi = 0) = 1 , for some unknown parameter . Using Uniform[0, 1] as the prior distribution for , find the posterior distribution. [Note: For your answer, and in answer to parts (b) and (d), give either a named distribution with its parameters, or a normalised density function.] [3 marks] I have collected a dataset of images, and employed an Amazon Mechanical Turk worker to label them. The labels are binary, nice or nasty. To assess how accurate the worker is, I first picked 30 validation images at random, found the true label myself, and compared the worker's label. The worker was correct on 25 and incorrect on 5. (b) Let be the probability that the worker labels an image incorrectly. Using Beta(0.1, 0.5) as the prior distribution for , find the posterior. [3 marks] I next ask the worker to label a new test image, and they tell me the image is nice. Let z {nice, nasty} be the true label, and let the prior distribution for z be Pr(nice) = 0.1, Pr(nasty) = 0.9. (c) For both z = nice and z = nasty, find P

worker says nice | z, . Hence find the posterior distribution of (z, ). Your answer may be left as an un-normalised density function. [5 marks] (d) Find the posterior distribution of z. [5 marks] My colleague has more grant money and she can employ 3 workers to rate each image. On a test set of 30 images, she found that they all agreed on 15 images, worker 1 was the odd one out on 8 of the images, worker 2 was the odd one out on 4, and worker 3 was the odd one out on 3. (e) Let i be the probability that worker i labels an image incorrectly. Find the posterior distribution of (1, 2, 3). Your answer may be left as an un-normalised density function. [4 marks] Hint. The Beta(, ) distribution has mean /( + ) and density Pr(x) = + 1 1 x 1 (1 x) 1 , x [0, 1] . Assuming you get a job as a network engineer in a multinational company that has five (5) regional station that must be interconnected with others for smooth operation of the organization. The company is about to network the regional offices together. Each physical link must allow communication in both directions. Use the knowledge acquired in this course to advice the management of the company based on the following: a) Recommend the most suitable Network topology for the organization. b) Give detail explanation of the recommended Topology. c) Illustrate the explanation in (b) with a diagram to show the interconnectivity of the five (5) regional offices. d) Explain four (4) major advantages of the topology named in (a) over other network topology

1. Enumerates five components of a data communications system. 2. Define key element of protocol. 3. Define two types of standards. 4. Describe the role of the following standards creation committee. 46 i. International Organization for Standardization (ISO). ii. International Telecommunication Union iii. American National Standards Institute (ANSI). iv. Institute of Electrical and Electronics Engineers (IEEE). 1) What are the three criteria necessary for an effective and efficient network? 2) Categorize the four basic topologies in terms of line configuration. 3) Name the four basic network topologies, and cite an advantage of each type. 4) For n devices in a network, what is the number of cable links required for a mesh, ring, bus, and star topology? 5) What are some of the factors that determine whether a communication system is a LAN or WAN?

Define data and Data Communication. 2. Compare analog and digital data 3. Why data communication 4. List and explain components of data communication system. 5. Highlights any 5 examples of resources that can be share on data communication and networks 6. Define and gives example of basic components of data communication network. 7. What are the major criteria that data communication network must meet. 23 8. Highlights the factors that affect response time as related to performance of data communication network. 9. List various ways of data representation in computer system. 10. Define Data communication equipment and data terminal equipment. Give at least two examples in each case.

What are the concerns of the physical layer in the Internet model? 2. What are the responsibilities of the data link layer in the Internet model? 3. What are the responsibilities of the network layer in the Internet model? 4. What are the responsibilities of the transport layer in the Internet model? 5. What is the difference between a port address, a logical address, and a physical address? 6. Name some services provided by the application layer in the Internet model. 7. How do the layers of the Internet model correlate to the layers of the OSI model? 8. How are OSI and ISO related to each other? 9. Match the following to one or more layers of the OSI model: a) Route determination b) Flow control c) Interface to transmission media d) Provides access for the end user 9. Match the following to one or more layers of the OSI model: a) Reliable process-to-process message delivery b) Route selection c) Defines frames d) Provides user services such as e-mail and file transfer e) Transmission of bit stream across physical medium 10. Match the following to one or more layers of the OSl model: a) Communicates directly with user's application program 55 b)Error correction and retransmission c) Mechanical, electrical, and functional interface d)Responsibility for carrying frames between adjacent nodes

Define communication software 2. W(a) The ACID properties are frequently used to characterize value-based semantics. (I) Define "atomicity" as utilized in the ACID setting. [1 mark] (ii) Define "toughness" as utilized in the ACID setting. [1 mark] (b) Write-ahead logging is a usually utilized plan to achieve value-based semantics while putting away an information base on a square stockpiling gadget, for example, a hard plate. (I) Under what conditions, during compose ahead log recuperation, can a exchange in the UNDO list be moved to the REDO list? [2 marks] (ii) Synchronously flushing submit records to circle is costly. How might we securely lessen coordinated I/O procedure on a high-throughput framework without forfeiting ACID properties? [2 marks] (iii) Describe two execution changes that could emerge from utilizing your answer to part (b)(ii). [2 marks] (c) (I) Transaction records in a compose ahead logging plan contain five fields: hTransactionID, ObjID, Operation, OldValue, NewValuei, however putting away the complete old and new qualities can consume huge measures of room. One methodology that may be utilized, for reversible activities being applied to certain information like XOR by a steady, is to store just the consistent contentions, instead of the full when information. What issues could happen because of this plan decision? (ii) Write-ahead logging frameworks should know the real on-plate area size for the compose ahead log to accurately act. A deviant plate merchant chooses to rebrand its 512-byte area circles as 2K-area plates, and changes the worth announced back to the information base framework. How should this influence data set uprightness? [4 marks] (iii) Explain how an information base seller who knows about the issue portrayed in part (c)(ii) moderate this issue in programming, and what impediments may there be to this methodology. [4 marks] 9 (TURN OVER) CST.2013.5.10 9 Concurrent and Distributed Systems Sun's Network File System (NFS) is the standard conveyed document framework utilized with UNIX, and has gone through a movement of variants (2, 3, 4) that have slowly further developed execution and semantics. (a) Remote technique call (RPC) (I) Explain how Sun RPC handle byte request (endianness). [2 marks] (ii) This approach might bring about pointless work. State when this happens and how should this be stayed away from. [2 marks] (b) Network File System variant 2 (NFSv2) and rendition 3 (NFSv3) (I) A key plan premise for NFS was that the server be "stateless" with deference to the client. State how this affects dispersed document securing in NFSv2 also, NFSv3. [2 marks] (ii) Another key plan premise for NFSv2 was the "idempotence" of RPCs; what's the significance here? [2 marks] (iii) One vital improvement in NFSv3 was the expansion of the READDIRPLUS RPC. Make sense of for what reason did this helps execution. [4 marks] (iv) NFSv3 carries out what is named "near open consistency" for document information reserving: assuming that client C1 keeps in touch with a record, shuts the document, and client C2 now opens the record for read, then, at that point, it should see the consequences of all composes gave by C1 preceding close. Notwithstanding, assuming C2 opens the document before C1 has shut it, then C2 might see some, all, or none of the composes gave by C1 (and in inconsistent request). Near open consistency is accomplished through cautious use of coordinated RPC semantics, joined with document timestamp data piggybacked onto server answers on all RPCs that work on records. Make sense of how near open consistency permits execution to be gotten to the next level. [4 marks] (v) NFSv3 adds another RPC, ACCESS, permitting the client to assign access control checks at document open opportunity to the server, as opposed to performing them on the client. This permits client and server security models to vary. Make sense of how this expansion additionally assists execute with shutting to-open consistency within the sight of perused storing.You are worried about decompilation from the objective directions of this machine back into a significant level programming language. (a) (I) What issues are raised by the condition codes, and what impact would they have on a nave decompilation? [5 marks] (ii) What enhancement methods would you be able to use to further develop matters, and consider the possibility that (any) are the impediments of your methodology. [5 marks] (b) Direct decompilation would have to prefix each memory access with a test to see whether the location being utilized was appropriately adjusted. This would prompt massive code that was difficult to peruse and had terrible showing. It is normal that, in the program that is being decompiled, unaligned memory gets to are exceptionally interesting. You are supposed to decompile into compact code in the undeniable level language that may not depend on any treatment of unaligned addresses on the new PC you arrange for. Once more, what streamlining strategies could you apply? [10 marks] 9 (TURN OVER) CST.2010.9.10 11 Quantum Computing (a) Which of coming up next are potential conditions of a qubit? (I) 1 2 (|1i + |0i) (ii) 0.6|0i + 0.4|1i (iii) 0.8|0i 0.6e 3i/4 |1i (iv) 3 2 i|0i 1 2 |1i [1 imprint each] For each legitimate state among the abovementioned, give the probabilities of noticing |0i what's more, |1i when the framework is estimated in the standard computational premise. [4 marks] (b) Suppose a two-qubit framework is in the state 0.8|00i + 0.6|11i. A Pauli X entryway (for example a NOT entryway) is applied to the second qubit, and a Hadamard door is applied to the first qubit. (I) What is the new condition of the framework? [2 marks] (ii) What are the probabilities of the potential results if both qubits are presently estimated? [2 marks] (c) Suppose we have a calculation which, given a blackbox registering an occasional work f with range {0, . . . , N 1}, decides the time of f. In addition, the calculation runs on schedule (log N) 2 . Make sense of how this would empower us to have a proficient (for example polynomial-time) calculation for considering numbers. [8 m

: Compose Java program that peruses a 1D exhibit of whole numbers from standard information and brings it back. a program that understands digits and creates them into numbers.

Talk about why the utilization of divided seeds in grouping information base pursuit is better compared to the utilization of successive seeds. [5 marks] (b) Discuss the intricacy of Sankoff's stinginess strategy. [5 marks] (c) Describe the four focuses conditions in phylogeny. [5 marks] (d) Discuss the suppositions of the Gillespie calculations. [5 marks] 3 (TURN OVER) CST.2010.9.4 4 Computer Vision (a) Consider an item's surface reflectance map (i, e, g) determining the sum of episode light reflected towards a camera from each point on a superficial level, where the point of the illuminant (a point source) comparative with the neighborhood surface ordinary N is I, the point comparative with N of a beam of light re-transmitted from the surface is e, and the point between the produced beam and the illuminant is g. (I) For what sort of surface is the reflectance map just (i, e, g) = cos(i)? Name this sort of surface and depict its key properties. [3 marks] (ii) For what sort of surface does the reflectance map rearrange to (i, e, g) = 1 on the off chance that I = e and both I and e are co-planar with the surface ordinary N, and (I, e, g) = 0 in any case? Name this sort of surface and depict its critical propertie