Advanced Topics in Computer Graphics
Advanced Topics in Computer Graphics CMPS 290
Popular in Course
Popular in ComputerScienence
This 19 page Class Notes was uploaded by Dr. Elyssa Ratke on Monday September 7, 2015. The Class Notes belongs to CMPS 290 at University of California - Santa Cruz taught by Staff in Fall. Since its upload, it has received 23 views. For similar materials see /class/182267/cmps-290-university-of-california-santa-cruz in ComputerScienence at University of California - Santa Cruz.
Reviews for Advanced Topics in Computer Graphics
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 09/07/15
Higher Order Polymorphism CMPS 290G Spring 2005 Presenter Kenneth Knowles May 26 2005 We motivated our discussion of higher order polymorphism by encoding the pair function using the kinds construct discussed earlier This looked as follows AX AY VZ X A Y A Z A Z A concrete example of this expression was given as the following PAIR NAT BOOL VZ NAT A BOOL A Z A Z Proceeding onward the cons and car expressions and their types were de ned as follows cons VXVYX A Y A PAIR X Y AX AY Az XAy YAZ Af X A Y A Zfzy car VXVYPAIR X Y A X AX AY Ap PAIR X YpXAz XAy Yr From these examples we notice that the following has been added to our syntax T VX KT V AX Ki t tT l AX Ki To summarize we have the following table which describes what we are able to map between including one that is foreshadowing of what is to come the representation of that mapping and the symbol we use to express that mapping Mapping Representation Symbol terms to terms Az Ti H types to terms AX KT types to terms AX Ki V terms to types Az TT Hz TT With our mappings de ned we produced a graphical representation of a few examples What resulted is the following illustration shown in gure I We concluded by re examining the A cube lling in all but two corners Thus we arrived at the A cube shown in gure 2 Direct to eBay Related Work Josh McCoy and Gillian Smith Although our prototype is to our knowledge the rst complete mobile commerce system designed for sellers in the developing world there is much supporting work that has inspired our design decisions 1 E Commerce in the Developing World Although the work in this paper addresses technical issues the larger goal of this project is to make a useable and viable system that works for the seller buyer and all others involved in the process As a consequence a signi cant portion of the work that will go into making this project a reality depends on the state of ECommerce in the developing world A report commissioned by the DFID Batchelor 2002 is optimistic about E commerce in the developing world for companies but does not address individual to individual commerce in detail The Third World Craft Development Center Third World Craft provides a middle layer consisting of services like an online store and quality assurance in crafted goods but does offer direct consumer to crafter transaction options Thinyane et a1 Thinyane explore how an intervention of information and communication technology can affect the Ecommerce of neglected rural areas This work leaves us optimistic about the success of our prototype 2 Mobile Commerce Mobile commerce or mcommerce is growing in popularity as portable devices become more prevalent Atkins et a1 Atkins et al discuss the possibility of extending existing ebusinesses by enabling telephones to access web databases eBay itself offers eBay Mobile eBay Mobile a mechanism for monitoring existing auctions and bidding on items from a cell phone However existing services and research focus on the buyer s experience rather than using mobile devices to sell items 3 Mobile Phone Use in the Developing World Mobile phones are increasingly common in developing nations despite an otherwise underdeveloped technical infrastructure Additionally work by Zhu et a1 Zhu et al shows that customers in the developing world are considered more likely to be early adopters of mobile technology People in these countries are welcoming mobile phones as lifechanging devices The fishing industry in India has greatly benefited from mobile phones fishermen can now use their cell phones to call different markets to determine the best price available for their catch Abraham In his work examining mobile phone usage in Cote d Ivoire Kamga Kamga tells the story of Mariama a hairstylist whose clientele used to be restricted to her family and close friends Her mobile phone provides a means of immediate contact and eases scheduling She now has many more customers and some are even far outside her village Such stories are now typical in these underdeveloped countries Horst et al The situation has led to a number of projects that use mobile phones to help solve social problems in the third world Gamos a small British company working focused on social issues in developing nations are working on using mobile phones for making payments and reducing costs of international money transfer Batchelor 2006 The MILLEE Mobile and Immersive Learning for Literacy in Emerging Economies proj ect Kam et al based at UC Berkeley attempts to increase literacy in school aged children in rural India The CAM mobile document processing system Parikh et al uses images captured from a camera phone as data entry providing the backups and the ability to analyze data from paperbased information processes We see the interest in and early successes of these projects as inspiration and encouragement for deploying our prototype in developing countries References Abraham R 2006 Mobile Phones and Economic Development Evidence from the Fishing Industry in India In Proceedings of the International Conference on Information and C 39 iun T 39 39 39 and D 39 I Berkeley California May 2006 ICTD 06 pp 4856 Atkins A S Ali A H and Shah H 2006 Extending ebusiness applications using mobile technology In Proceedings of the 3rd international Conference on Mobile Technology Applications ampAmp Systems Bangkok Thailand October 25 27 2006 Mobility 3906 vol 270 ACM Press New York NY 44 DOI httpdoiacmorg10 11451292331 1292381 Batchelor SJ Webb M ECommerce Options for Third World Craft Producers Technical Report DFID Knowledge and Research Project R7792 March 2002 Batchelor SJ Briefing Note Why MBanking and MPayments is Potentially Transformational 2006 httpwwwgamosorguldmpayments eBay Mobile httppagesebaycommobile retrieved October 22 2007 Horst H and Miller D 2006 The Cell Phone An Anthropology of Communication Berg Publishers Kam M Ramachandran D Raghavan A Chiu J Sahni U and Canny J Practical Considerations for Participatory Design with Rural School Children in Underdeveloped Regions Early Re ections from the Field In Proceedings of the ACM Conference on Interaction Design and Children Tampere Finland June 2006 IDC 06 ACM Press New York NY Kamga O 2006 Mobile Phone in Cote d Ivoire Uses and SelfFulfillment In Proceedings of the International Conference on Information and Communication Technologies and Development Berkeley California May 2006 ICTD 06 pp 184 192 Early measurements from the Internet Archive November 2002 1 Introduction The Archive recently moved its Webcrawl data from tape to disk allowing computational studies of the material So far we have undertaken only a few very simple measurements the results of which are reported herein 2 Crawl cataloging There is currently no catalog of the Archive s Web data That is there is no way of identifying the crawls to which the various page instances belong nor the crawl policies that de ned those crawls An important goal in analyzing this data is distinguishing crawling artifacts from true trends in the underlying Web As we will see without a catalog this is impossible to do Starting in 2001 the Archive started doing a better job segregating its crawl data In particular it started pref1xing all data from the same crawl with a twoletter code The first such crawl used the code DD the next DE and so forth There is no policy data on even these crawls however And prior to the institution of this convention there is no way to segregate crawl data by crawl We made two attempts at segregating this older data neither of which was successful The first attempt was to create a histogram of instance counts by week We were expecting to see a cyclical pattern of many heavy weeks separated by a week or two of downtime between crawls Unfortunately no such pattern existed this same data aggregated by month is presented in the next Section Preliminary investigation suggests that this failure is due to the fact that crawls were overlapped in time The second attempt was based on examination of fllename patterns We were hoping that we could identify informal conventions that might provide hints We found over 50 patterns The table on the next page shows a little over 25 of these the ones covering the largest number of files On the one hand fllename patterns are clearly not going to directly give us the catalog we are looking for On the other hand these patterns do appear to be useful and we intend to investigate them further Some preliminary investigations have led to the following findings Several patterns start with the prefixes arc or crc Arc bro and crc were the names of machine clusters used at Alexa for storing crawl data Alexa used to keep an old crawl on two of these for data mining while using the third for capturing the next crawl Thus it s likely that the arc and crc files contain one or two crawls each Sarah and green were likewise names of machines at Alexa we do not yet know the conventions under which those machines were used The IA les are verv eaer les Unfortunatelv there is reason to believe that in this Regular Expression Pattern les quotlA096X arc gz A000001arcgz 15736 l FS 096arcgz FS000004arcgz 69590 G R096X39 arc gz GR000086 arc 92 101099 quotgreenO093199901209891097arcgz green00001 99901 2700503291 7427830 arc 92 109834 quotgreenO13093199903 509892097arcgz green00001 999041 6202859 92431 9739 arc 92 124347 quotgreenOO01093199908 91009893097X02arcgz green0000001 999081 91 32249935095489 arc 92 150246 green0001093199910220000109894097Xarcgz green0000001 9991 1021 53301 941 586794 arc 92 178155 quotgreenOO01 0 93 200002 5 09895097 arc gz green000000200002281 8521 7951 793472 arc 92 216574 quotgreenOO01 0932000056 09896097 arc gz green00024820000529233819 960000081 arc 92 232034 quotgreenO0 9391097arcgz green000091 24831 98 arc 92 235541 l sz1r31h00951 9991 0098930 97X39 arc gz sarah001 8001 9991 0071 60426 939337644 arc 92 239019 l sz1rah0095199911200001209894097arcgz sarah00000019991102145335941583381arcgz 257450 l sz1rah0095200002 5 09895097 arc gz sarah00000020000225152701951521513arcgz 299828 quotsarah009520000609896097arcgz sarah001 000 2000060222051 7 960009065 arc 92 319806 quot20000701060009420000709896097arcg 20000706000000200007071 42900 963005896 arc 92 324609 quot200007100000250920000711209696097arcgz 20000710000020 20000711140102 963391175arcgz 325217 quot20000929101618240000932000091097097arcgz 20000929000000 200009291 74801 970278439 arc 92 377266 quotaug0912Xarcg aug000814222632arcgz 11086 quotcrc12247100925aag956097arcgz crc127ae957919701arcgz 419570 quotcr0204200011098arcgz crc2020001109142600arcgz 433480 l 31rc0909200012098arcgz arc2820001210123241arcgz 487739 quotarc09 0 9 2001 01 020 98 arc gz arc1320010225073701arcgz 543344 l DD51rlt12092001024098v151rdgz DDarc1120010301063942arcgz 661073 quotDE arc1209crawl16200103 6098arcgz DE arc1820010330010418arcgz 710984 quotDF crawl2 82001058098arcgz DF crew1220010602020727arcgz 799708 l DGltr31111820010791009831rcgz DGcrawi120010803233813arcgz 888230 l DHalexa023cr31wl18200109102098arcgz DHaexa020011120030016arcgz 943601 3 Collectlon rates The following table summarizes the rate at which data has come into the Archive over the past siX years A E 1 000 1 00 V E 800 80 3 O 600 60 0 g 400 40 C as 200 20 a E 0 Jan96 Jan97 Jan98 Jan99 Jan00 Jan01 Jan02 The red lines indicate the number of HTML instances acquired per month the blue lines indicate the total number of instances per month the green triangles plot the ratio of the two This data starts in Feb 1996 During 1996 the incoming rate was so low that it doesn t register on this graph For the entire year of 1996 355M instances were obtained our monthly rate surpassed this by March 1997 Recently the monthly acquisition rate has been close to a billion pages per month This graph is clearly in uenced much more heavily by crawling artifacts than by underlying trends in the Web Certainly the Web was growing quickly in that period of time and the growth in collection rate was driven by that underlying growth But the growth in collection rate and the temporary setbacks in that growth was dictated as much by storage and bandwidth availability by crawlingsoftware limitations and by operational glitches The ratio of HTML to other media types is even more so an artifact of crawling policy It must be understood that most of this data was collected by Alexa Internet whose primary goal was maintaining it s Internet Search service rather than archiving the Web It turns out that HTML was really the only media they needed to download to run their server Thus it is not surprising that at times practical considerations forced Alexa to abandon crawling media other than HTML In fact anecdotal evidence suggests that other Web Search companies have ip opped on the crawling of nonHTML data and also that most of them did not crawl nonHTML during 1999 4 Instance sizes The following graph gives the mean instance size averaged per month since 1996 Again red lines are for HTMLonly instances blue lines are for all instances Mean instance size KB 1 11111111 1111 111MWHHWHHHIHHHH Because instancesize is heavily dependent on media type and because of the inconsistency of crawl policy regarding collection of nonHTML data the blue lines are not very informative However the red lines indicate a clear upward trend in the size of Inferring Phylogenetically Conserved Motifs Bernard Suh Department of Biomolecular Engineering University of California Santa Cruz Suh CMF S 290C Project Talk 060904 p111 Biological Motivation A pramorer cxsvreguatory region rranscrr pnon um muduxa basal core enhancer promoter 9 quot quotquot quot TR B chromatin remodellng compex TAPS lranscription coJactols pol halaenzyme r transcrrpn on factors transcnpnon sran site V r TATA box TATArbindmg prorem Lquot oopmg lactors chlomaun modules chromatin Figure 1 Wray et al 2003 VCMPS zanc Pvujecl Tam Demn4 7 p 211 E Motif Search Problem Given n sequences of length l with alphabet ACGT 0 Given a tree topology with 72 leaves 0 Star 0 Binary 39 Specify motif length A 39 Assume each sequence contains 1 motif 0 Define backgroundnull model 39 Given a method of scoring motifs 0 Given an optimality criterion relative entropy 0 Find the position d within each sequence of the best motif Consensus motifpiltAgtp2lt0gtpiGuam Suh CMF S 290C Project Talk 060904 p311 Gibbs Sampler Lawrence et al 1993 I 4MIHM2HM3HM4HMSHM6F a I bg I 0 HMM Representation vs Scanning 0 Position d of motif in sequence is hidden 0 Learning via Stochastic EM Assumes startree I I Suh CMPS 2900 Project Talk 060904 p411 FootPrinter Blanchette et al 2000 0 Standard Parsimony 0 Match 0 O Mismatch 1 MAX PRODUCT with logs Viterbi 39 Time complexity Onk4k2 Suh CMF S 290C Project Talk 060904 p511 Probabilistic Models for DNA JukesCantor 1969 JC69 extension F81 9 999 999 9 Kimura 2 parameter 1980 K2P or K80 extension HKY85 T139Tlgt El I71 T139713 Suh CMF S 290C Project Talk 060904 p611 Hybrid Approach for Motif Search 39 Combine approaches from Gibbs Sampler and FootPrinter 0 Enhance FootPrinter with SUM PRODUCT 0 Probabilistic model is a phyloHMM 0 Replace emission prob with tree likelihood prob O Felsenstein amp Churchill 1996 Siepel amp Haussler 2003 MCMC via Metropolis updates for parameters 0 6 6N01 Suh CMF S 290C Project Talk 060904 p711 Synthetic Data generate 72 0th order Markov background sequences of length l 0 Given a tree with 71 tips evolve a motif of length A t Implant motif into background sequence Suh CMF S 290C Project Talk 060904 p811 Sample Results 40 45 4gt o a l l l l l l l l 0287 8 Q 00207 o a O 00247 Q 037 00227 00257 0027 0027 0018 0015 0010 001 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 p 0 5 i t i 0 n p 0 5 i t i 0n 0 n 4k 6l 50HKY85 Gibbs right Hybrid left 39 Summarize 0 PME Credible Interval O Consensus motif Suh CMF S 290C Project Talk 060904 p911 Sample Results continued 004 I I I I I I I I 0035 003 Q 9 Q g 0025 3 I O Q 002 0015 001 0 5 10 15 20 25 30 35 40 45 position O 2 motifs implanted Suh CMF S 290C Project Talk 060904 p1011 TNATTOPT xTT gtTxuxTlzzTi lt1sz gt VFOLDTv NATLIST 711 lt NIL UNIT CONS NAT X gt UNFLDUXT 5 gt uxTT FOLD does opposite example of unfolding NATLIST once ltNIL UNIT CONS NAT ltNILUNIT CONS NAT X gt gt Example of type rule for a pair I E t T I E 152 T2 TEtt2T gtltT2 Type rules involving FOLD and UNFOLD t gt t UNFOLWT FOLDSv gt v T Ft uxT U r gt uxTT r r UNFOLDMLTV U T F 15 U U r gt uxTT r r FOLDTt WT A suggestion for cons mknil UnitFOLDNATLIST ltnil unitgt mknil Unit gtNatlist mkcons Nat gtlt Natlist gt Natlist Greatest xed point Fixed point is anywhere TRR now R is a set greatest xed point is largest set R for which this is true smallest xed point is smallest set R for which this is true To get least xed point start out with empty set and add To get greatest xed point start with all sets and trim Q is there any in nite tree in the least xed pointI gt at the limit THE which is the last set in the xed point if this has an in nite tree then all the above also have that in nite tree so the set cannot be nite so it cannot be a least xed point The greatest xed point is very different from the least xed point makes point that Prolog looks for the greatest xed point Q is do I want to start at null and say or start at all and say no F Ptype gtlt type gt Ptype gtlt type FS UTOP NatNat U 7 UU gtlt U2 T x T2 UT E S U2 T2 6 S UU gt U2 T gt T2 1 U T e s U212 e S UU xT U x gt mmTl E S l139uxU T r gt uxlflUT ES U Conjecture from the board uXX uXX E LFWF follows our algorithm to conclude that even this simple type will not check with least xed point uXNat gtlt XNat gtlt uXNat x X E F quotEI Nat gtlt uXNat x X E F 2EI uXNat x X E Fit 35 Bounded Quanti ed Types Types and Programming Languages7 Spring 20057 SoE UCSC PRESENTER Jessica Gronski SCRIBE Avik Chaudhuri May 19 2005 1 Looking back We recalled the differences in programming styles when using objects versus ADTs7 in particular the expressiveness problem with objects under pure existential types when de ning strong binary operations Such questions came up in Kim Bruce s talk on May 17 as well We agreed that more information on the type variable in existential types is useful in order to write interesting programs bounded existentials provide the means to pass such information 2 On the board The syntax of the pure lambda calculus with bounded quanti ed types has the following syntax erms t z varia e l z Tt abstraction l t 25 application l AX lt Tt type abstraction l t T type application l T7 25 as T packing l let X7 z t in t unpacking Types T X type variable l Top maximum type l T a T function type l VXlt T T universal type l 3Xlt T7 T existential type Environments P 0 empty l P z T variable type declaration l P Xlt T type bound declaration 1
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'