SAO: A Stream Index for Answering Linear Optimization Queries

Size: px
Start display at page:

Download "SAO: A Stream Index for Answering Linear Optimization Queries"

Transcription

1 SAO: A Stream Index for Answerng near Optmzaton Queres Gang uo Kun-ung Wu Phlp S. Yu IBM T.J. Watson Research Center {luog, klwu, psyu}@us.bm.com Abstract near optmzaton queres retreve the top-k tuples n a sldng wndow of a data stream that maxmze/mnmze the lnearly weghted sums of certan attrbute values. To effcently answer such queres aganst a large relaton, an onon ndex was prevously proposed to properly organze all the tuples n the relaton. However, such an onon ndex does not work n a streamng envronment due to fast tuple arrval rate and lmted memory. In ths paper, we propose a SAO ndex to approxmately answer arbtrary lnear optmzaton queres aganst a data stream. It uses a small amount of memory to effcently keep track of the most mportant tuples n a sldng wndow of a data stream. The ndex mantenance cost s small because the great majorty of the ncomng tuples do not cause any changes to the ndex and are quckly dscarded. At any tme, for any lnear optmzaton query, we can retreve from the SAO ndex the approxmate top-k tuples n the sldng wndow almost nstantly. The larger the amount of avalable memory, the better the qualty of the answers s. More mportantly, for a gven amount of memory, the qualty of the answers can be further mproved by dynamcally allocatng a larger porton of the memory to the outer layers of the SAO ndex. We evaluate the effectveness of ths SAO ndex through a prototype mplementaton.. Introducton Data stream applcatons are becomng popular. Many of such applcatons use varous lnear optmzaton queres [2, 3, 4] to retreve the (approxmate) top-k tuples that maxmze or mnmze the lnearly weghted sums of certan attrbute values. For example, n envronmental epdemologcal applcatons, varous lnear models that ncorporate remotely sensed mages, weather nformaton, and demographc nformaton are used to predct the outbreak of certan envronmental epdemc dseases, lke Hantavrus Pulmonary Syndrome [3]. In ol/gas exploraton applcatons, lnear models that ncorporate drll sensor measurements and sesmc nformaton are used to gude the drllng drecton [4]. In fnancal applcatons, lnear models that ncorporate personal credt hstory, ncome level, and employment hstory are used to evaluate credt rsks for loan approvals [3]. In all the above applcatons, data contnuously stream n (say, from satelltes and sensors) at a rapd rate. Users frequently pose lnear optmzaton queres and want answers back as soon as possble. Moreover, dfferent ndvduals may pose queres that have dvergent weghts and K s. Ths s because the optmal weghts may vary from one locaton to another (n ol/gas exploraton), the weghts may be adjusted as the model s contnually traned wth hstorcal data collected more recently (n envronmental epdemology and fnance), and dfferent users may have dfferng preferences. In a read-mostly envronment, Chang et al. [2] frst proposed an onon ndex to speed up the evaluaton of lnear optmzaton queres aganst a large database relaton. An onon ndex organzes all the tuples n the database relaton nto one or more convex layers, where each convex layer s a convex hull. For each, the ( + )th convex layer s contaned wthn the th convex layer. For any lnear optmzaton query, to fnd the top-k tuples, we need to search no more than all the vertces of the frst K outer convex layers n the onon ndex. However, due to the extremely hgh cost of computng precse convex hulls, both the creaton and the mantenance of the onon ndex are rather expensve. Moreover, an onon ndex requres lots of storage because t keeps track of all the tuples n a relaton. In a streamng envronment, tuples keep arrvng rapdly whle avalable memory s lmted. Hence, t s mpossble to mantan a precse onon ndex for a data stream, let alone usng t to provde exact answers to lnear optmzaton queres. To address these problems, we propose a SAO (Stream Approxmate Onon-lke structure) ndex for a data stream. The ndex provdes hgh-qualty, approxmate answers to arbtrary lnear optmzaton queres almost nstantly. Our key observaton s that the precse onon ndex typcally contans a large number of convex layers, but most nner layers are not needed for answerng lnear optmzaton queres. Hence, the SAO ndex mantans only the frst few outer convex layers. Moreover, each layer n the SAO ndex only keeps some of the most mportant vertces rather than all the vertces. As a result, the amortzed mantenance cost of a SAO ndex s rather small because the great majorty of the ncomng tuples, more than 95% n most cases, do not cause any changes to the ndex and are quckly dscarded, even though ndvdual nserts or deletes mght have non-trval costs. A key challenge n desgnng a SAO ndex s: For a gven amount of memory, how do we properly allocate t among the layers so that the qualty of the answers can be maxmzed? To do that, a dynamc, error-mnmzng storage allocaton strategy s used so that a larger porton of the avalable memory tends to be allocated to the outer layers than to the nner layers. In ths way, both storage and mantenance overheads of the SAO ndex are greatly reduced. More mportantly, the errors ntroduced nto the approxmate answers are also mnmzed.

2 Wth lmted memory and contnually arrvng tuples, there are ntrnsc errors n any stream applcaton. It s dffcult to provde an upper bound on these errors for lnear optmzaton queres because the amount of naccuraces depends on the specfc sequence of tuples n a stream. However, n practce, the exact errors can be measured based on stream traces. As shown n the experments conducted n ths paper, the actual errors are relatvely mnor (often less than %) even f the SAO ndex holds only a tny fracton (less than 0.%) of the tuples n the sldng wndow. Ths s because, statstcally, only few tuples cause errors. Moreover, the mpact of any error, no matter how large t may be, dsappears mmedately once the tuple causng the error has moved out of the sldng wndow. For some stream applcatons, the lnear optmzaton queres are known n advance and the entre hstory, not just a sldng wndow, of the stream s consdered. In ths case, for each query, an n-memory materalzed vew can be mantaned to contnuously keep track of the top-k tuples. However, f there are many such queres, t may not be feasble to keep all these materalzed vews n memory and/or to mantan them n real tme. As a consequence, the SAO ndex method s stll needed under such crcumstances. We mplemented the SAO ndex by modfyng a wdelyused Qhull package []. Our expermental results show that the SAO ndex can handle hgh tuple arrval rates, be mantaned effcently n real tme, and provde hghqualty answers to lnear optmzaton queres almost nstantly. 2. Revew of the Tradtonal Onon Index We brefly revew the earler onon ndex [2] for lnear optmzaton queres aganst a large database relaton. Suppose each tuple contans n numercal feature attrbutes and m 0 other non-feature attrbutes. A top-k lnear optmzaton query asks for the top-k tuples that maxmze the followng lnear equaton: n j max{ w } a, top K = where j j j ( a, a2,..., an ) s the feature attrbute vector of the jth tuple and ( w, w2,..., wn ) s the weghtng vector of the query. Some w s may be zero. Here, n j v j = w = a s called the lnear combnaton value of the jth tuple. A set of tuples S can be mapped to a set of ponts n an n-dmensonal space accordng to ther feature attrbute vectors. For a top-k lnear optmzaton query, the top-k tuples are those K tuples wth the largest projecton values along the query drecton. The onon ndex n [2] organzes all the tuples nto one or more convex layers. The frst convex layer s the convex hull of all the tuples n S. The vertces of form a set S S. For each >, the th convex layer s the convex hull of all the tuples n form a set S U S S j = j S U S j = j. The vertces of. It s easy to see that for each, + s contaned wthn. Accordng to lnear programmng theory, we have the followng property [2]: Property : For any lnear optmzaton query, suppose all the tuples are sorted n descendng order of ther lnear combnaton values (v j ). The tuple that s ranked kth n the sorted lst s called the kth largest tuple. Then the largest tuple s on. The second largest tuple s on ether or 2. In general, for any, the th largest tuple s on one of the frst outer convex layers. Gven a top-k lnear optmzaton query, the search procedure of the onon ndex starts from and searches the convex layers one by one. On each convex layer, all ts vertces are checked. Based on Property, the search procedure can fnd the top-k tuples by searchng no more than the frst K outer convex layers. 3. SAO Index The orgnal onon ndex [2] keeps track of all the tuples, requrng lots of storage. Mantanng the orgnal onon ndex s also computatonally costly, makng t dffcult to meet the real-tme requrement of data streams. To address these problems, we propose a SAO ndex for lnear optmzaton queres aganst a data stream. Our key dea s to reduce both the ndex storage and mantenance overheads by keepng only a subset of the tuples n a data stream n the SAO ndex. We focus on the count-based sldng wndow model for data streams, wth W denotng the sldng wndow sze. That s, the tuples under consderaton are the last W tuples that we have seen. Our technques can be easly extended to the case of tmebased sldng wndows or the case that the entre hstory of the stream s consdered. Suppose the avalable memory can hold M + tuples. In the steady state, no more than M tuples are kept n the SAO ndex. That s, the storage budget s M tuples. In a transton perod, M + tuples can be kept n the SAO ndex temporarly. Our technques can be extended to the case where memory s measured n bytes. In general, a tuple contans both feature and non-feature attrbutes. We are nterested n fndng all the attrbutes of the top-k tuples. Hence, all the attrbutes of those tuples n the SAO ndex are kept n memory. Even f the convex hull for feature attrbutes occupy only a small amount of space, the non-feature attrbutes may stll domnate the storage requrement. For example, n the earler-mentoned, envronmental epdemology applcaton, each tuple has a large non-feature mage attrbute, whch s also kept n memory. Note that the mage cannot be stored on dsk, even f we lke to do so, because the tuple arrval rate can

3 be too hgh for even the fastest dsk to keep up wth the rapdly arrvng tuples. Our desgn prncple s as follows. To provde hghqualty answers to lnear optmzaton queres, the SAO ndex carefully controls the number of tuples on each layer. It dynamcally allocates proper amount of storage to ndvdual layers so that a larger porton of the avalable memory tends to be allocated to the outer layers. As such, the qualty of the answers can be maxmzed wthout ncreasng the storage requrement. In case of overflow, the SAO ndex keeps the most mportant tuples and throws away the less mportant ones. Moreover, to mnmze the computaton overhead, the creaton and mantenance algorthms of the SAO ndex are optmzed. 3. Index Organzaton The SAO ndex s based on a key observaton: An onon ndex typcally contans a large number of convex layers, but most nner layers are not needed for answerng the majorty of lnear optmzaton queres. The SAO ndex keeps only the frst outer convex layers, where s specfed by the user creatng the ndex. Snce M s lmted, the SAO ndex cannot always keep the precse frst outer convex layers. Therefore, for each of the frst outer convex layers, the SAO ndex may only keep some of the most mportant tuples rather than all the tuples belongng to that layer. In other words, each layer n the SAO ndex s an approxmate convex layer (AC) n the sense that t s an approxmaton to the correspondng precse convex layer n the onon ndex. For each ( ), s used to denote the th AC. The SAO ndex mantans the followng propertes. Each AC s the convex hull of all the tuples on that layer. For each ( ), + s contaned wthn. Also, the total number of tuples on all ACs s no more than M. All the tuples n the SAO ndex are kept as a sorted, doubly-lnked lst dl. The sortng crteron s a tuple s remanng lfetme. The frst tuple n dl s gong to expre the soonest. In ths way, we can quckly check whether any tuple n the SAO ndex expres, whch s needed at Step 2 of Secton 3.5 below. Also, we can easly delete tuples that are n the mddle of dl, whch s necessary when the avalable memory s exhausted and a tuple needs to be deleted from the SAO ndex (see Secton 3.3 below). For each AC, a standard convex hull data structure s mantaned. The vertces of the convex hull pont to tuples n dl. Also, each tuple t n dl has a label ndcatng the AC to whch tuple t belongs. Ths label s used when a tuple expres and needs to be removed from the correspondng AC (see Secton 3.5 below). 3.2 Allocatng Proper Memory to Each ayer A key challenge n desgnng a SAO ndex s: For a gven amount of memory, how do we properly allocate the memory to each layer so that the qualty of the answers can be maxmzed? A Smple, Unform Storage Allocaton Strategy A smple storage allocaton strategy s to dvde the storage budget M evenly among all ACs. Each AC cannot keep more than M / tuples. However, ths smple, unform method s far from beng optmal. In order to provde hgh-qualty answers to lnear optmzaton queres, the SAO ndex should allocate more tuples to the outer ACs than to the nner ACs, as outer ACs contan the largest tuples. Statc, Error-Mnmzng Storage Allocaton Now we descrbe a statc, error-mnmzng storage allocaton strategy when resource s lmted. We determne the optmal numbers of tuples the SAO ndex should allocate to the ACs. By resource beng lmted, we mean that each AC needs more tuples than can be actually allocated to t. For each ( ), let N denote the optmal number of tuples that should be allocated to. Then N M =. () = Consder a top- lnear optmzaton query. For each ( ), let t represent the exact th largest tuple, and t ' represent the th largest tuple that s found n the SAO ndex. Here, v s the lnear combnaton value of t, and v ' s the lnear combnaton value of t '. The relatve error of t ' s defned as e = ( v v ) / v. (2) For the top- tuples (t ') that are returned by the SAO ndex, a weghted mean of ther relatve errors s used as the performance metrc e: e = u = e / u, (3) = where u s the weght of e. Intutvely, the hgher the rank of a tuple t, the more mportant t s relatve error. Hence, u should be a non-ncreasng functon of. We would lke to mnmze the mean of e for all top- lnear optmzaton queres, whch s how N s are derved. Our dea s to represent the mean of e as a functon of N s and fnd ts mnmal value under condton (). Accordng to our dervaton whose detals are n [5], we can show that N C, where C = j 2 = /. (4) j 3.3 Dynamc, Error-Mnmzng Storage Allocaton The real world s not statc. At any tme, some ACs may need more than N tuples whle other ACs may need fewer than N tuples. As tuples keep enterng and leavng the sldng wndow, the storage requrements of dfferent ACs change contnuously. To ensure the best qualty of the answers, the SAO ndex needs to fully utlze the storage budget M as much as possble by dong dynamc storage allocaton. Our desgn prncple s: Whenever possble, the storage budget M s used up. At the same tme, the SAO ndex tres ts best to mantan the condton that the number of tuples on s proportonal to C.

4 The concrete method s as follows. For each ( ), let M denote the number of tuples on. The SAO ndex contnuously montors these M s. At any tme, there are two possble cases. In the frst case, M M. Ths s = a normal case and nothng needs to be done, as the storage budget M has not been used up. In the second case, = M = + M. (Accordng to Secton 3.5, = M can never be larger than M +.) Ths s an overflow case, as the storage budget M s exceeded by one. We need to pck a canddate AC and delete one tuple from t. Choosng a Canddate Approxmate Convex ayer We frst dscuss how to choose the canddate AC. For each ( ), let r = M / N. We pck j such that rj = max{ r r >, }. Ths j must exst. Otherwse ( ), r. Ths leads to M N M =, = = whch conflcts wth the condton that = M = M +. j s chosen as the canddate AC. Choosng a Canddate Tuple Now one canddate tuple needs to be deleted from the canddate AC j. Intutvely, ths canddate tuple t should have a close neghbor so that deletng t wll have lttle mpact on the shape of j. Two tuples on an AC are neghbors f they are connected by an edge. For any tuple t on j, let R t denote the Eucldean dstance between tuple t and ts nearest neghbor on j. The canddate tuple s chosen to be the tuple that has the smallest R t (usually there are a par of such tuples and the older one,.e., the sooner-to-expre one, s pcked). Deletng Canddate Tuple Fnally, after choosng the canddate tuple t, we use the method that s descrbed n Step 2 of Secton 3.5 below to delete t from j and then adjust the affected ACs. 3.4 Index Creaton At the begnnng, the SAO ndex s empty. We keep recevng new tuples untl there are M tuples. Then a standard convex hull constructon algorthm s used to create the ACs n batch. From now on, each tme a new tuple arrves, we use the method n Secton 3.5 to ncrementally mantan the SAO ndex. 3.5 Index Mantenance In a typcal data streamng envronment, we expect that W >> M,.e., only a small fracton of all W tuples n the sldng wndow are stored n the SAO ndex. Intutvely, ths means that tuples on the ACs can be regarded as anomales. The smaller the ( ) s, the more anomalous the tuples on are. As a result, we have the followng heurstc (not exact) property: Property 2: Most new tuples are normal tuples and thus nsde. Moreover, for a new tuple t, t s most lkely to be nsde. ess lkely s tuple t between - and, and even less lkely s tuple t between -2 and -, etc. Accordng to our storage allocaton strategy descrbed n Sectons 3.2 and 3.3, the nner ACs tend to have fewer tuples than the outer ACs. From computatonal geometry lterature, t s known that gven a pont p, the complexty of checkng whether p s nsde a convex polytope P ncreases wth the number of vertces of P. Therefore, we have the followng property: Property 3: For a tuple t, t s typcally faster to check whether t s nsde an nner AC than to check whether t s nsde an outer AC. Upon the arrval of a new tuple t, Propertes 2 and 3 are used to reduce the SAO ndex mantenance overhead. We proceed n three steps. Step : Tuple Inserton All ACs are checked one by one, startng from. From Propertes 2 and 3 together wth the procedure descrbed below, t can be seen that ths checkng drecton s the most effcent one. There are two possble cases. In the frst case, tuple t s nsde. Accordng to Property 2, ths s the mostly lkely case. Also, accordng to Property 3, t can be dscovered quckly whether tuple t s nsde. In ths frst case, tuple t wll not change any of the ACs and thus can be thrown away mmedately. Snce no new tuple s ntroduced nto the SAO ndex, there wll be no memory overflow. Hence, Step 3 can be skpped, although Step 2 stll needs to be performed. Note: If s empty, we thnk that tuple t s outsde of. In the second case, a number k ( k ) can be located such that tuple t s nsde k- but outsde of k. (If k =, tuple t s outsde of all ACs.) In ths case, tuple t needs to be nserted nto the SAO ndex n a way smlar to that the onon ndex s mantaned [2]. Step 2: Tuple Expraton The arrval of tuple t wll cause at most one tuple n the SAO ndex to expre from the sldng wndow. et t' denote the frst tuple n the doubly-lnked lst dl, whch s the only tuple n the SAO ndex that may expre from the sldng wndow. There are two possble cases. In the frst case, tuple t' has not expred. We proceed to Step 3 drectly. In the second case, tuple t' has expred and thus needs to be deleted from the SAO ndex n a way smlar to that the onon ndex s mantaned [2]. Step 3: Handlng Memory Overflow In the above two steps, at most one new tuple s ntroduced nto the SAO ndex whle one or more tuples may be deleted (e.g., tuples may get expelled from n

5 = Step ). Now we check whether condton M M stll holds. If not, = M = + must be true. In ths M case, we use the procedure that s descrbed n Secton 3.3 to delete one tuple from the SAO ndex. 3.6 Query Evaluaton To provde approxmate answer to a top-k lnear optmzaton query (K can be larger than ), we use the onon ndex search procedure that s descrbed n [2]. 4. Performance Evaluaton We mplemented our technques by modfyng the wdely used Qhull (verson 2003.) software package [], whch mplements effcent constructs for the creaton of convex hulls. Our measurements were performed on a computer wth one.6ghz processor, GB man memory, one 75GB dsk, and runnng the Mcrosoft Wndows XP. Our evaluaton used the real data set n the 2005 UC data mnng competton [6]. Among all the attrbutes, we used the three (n=3) attrbutes that carred the most nformaton (.e., had the largest number of dstnct values) as the feature attrbutes. In each experment, after the system has run for enough tme and reaches a steady state, we posed 00 top-k lnear optmzaton queres, whose weghts were unformly dstrbuted between - and. As n Secton 3.2, for each top-k lnear optmzaton query, the weghted relatve K K error e s defned as e = u = e / u, where = u = / ( K ). The followng three performance metrcs were used: () Max error: The maxmum e observed for the 00 top-k lnear optmzaton queres. (2) Avg error: The average e observed for the 00 top-k lnear optmzaton queres. (3) Throughput: In the steady state, the average number of tuples that can be processed per second. In the rest of ths secton, by default the sldng wndow sze s W=,000,000. The SAO ndex contans =4 ACs. The number of top tuples s K=0. The storage budget M s 500 tuples. The dmensonalty n (.e., the number of feature attrbutes) of the data set s 3. We vared the storage budget M from 200 tuples to 500 tuples. We used the dynamc storage allocaton strategy descrbed n Secton 3.3 and compared the followng two methods of computng N s: () Error-mnmzng method: Ths s the method descrbed n Secton 3.2. N s are computed accordng to (4). (2) Unform method: N = M /. Fgure shows the mpact of M/W rato on the weghted relatve error. For any M/W rato, both the max error and the avg error are much smaller for the error-mnmzng method. Namely, the error-mnmzng method works better than the unform one. weghted relatve error throughput (tuples/second) 25% 20% 5% 0% 5% avg error, error-mnmzng max error, error-mnmzng avg error, unform max error, unform 0% 0.02% 0.03% 0.04% 0.05% M /W Fgure. Weghted relatve error vs. storage budget % 0.03% 0.04% 0.05% M /W Fgure 2. Throughput vs. storage budget. Both the max error and the avg error decrease as M/W rato ncreases. When M/W=0.05%, the avg error s 0.5% whle the max error s 3%, both farly small. In other words, even wth a storage budget that s only a very small fracton of the sldng wndow sze, the SAO ndex can provde farly accurate answers. Fgure 2 shows the mpact of M/W rato on the throughput. The y-axs uses logarthmc scale. When M/W=0.05%, the throughput s over 00,000 tuples/second. [5] s a full verson of ths paper that ncludes addtonal results and algorthm detals. In summary, n all our experments, usng the SAO ndex to answer a top-k lnear optmzaton query always takes less than second. Usng a farly small amount of memory space, the SAO ndex provdes good approxmate answers to top-k lnear optmzaton queres, and the qualty of approxmate answers mproves wth the amount of avalable memory. Even under a rapd tuple arrval rate, the SAO ndex can stll be mantaned effcently n real tme. Hence, the SAO ndex can provde good support for answerng lnear optmzaton queres aganst a data stream. References [] C.B. Barber, D.P. Dobkn, and H. Huhdanpaa. The Quckhull Algorthm for Convex Hulls. ACM Trans. Math. Softw. 22(4): , 996. [2] Y.C. Chang,.D. Bergman, and V. Castell et al. The Onon Technque: Indexng for near Optmzaton Queres. SIGMOD Conf. 2000: [3] C.S., Y.C. Chang, and.d. Bergman et al. Model-Based Mult-modal Informaton Retreval from arge Archves. ICDCS Internatonal Workshop of Knowledge Dscovery and Data Mnng n the World-Wde Web, [4] C.S., Y.C. Chang, and J.R. Smth et al. SPIRE/EPI-SPIRE Model-Based Mult-modal Informaton Retreval from arge Archves. MMCBIR 200. [5] G. uo, K. Wu, and P.S. Yu. SAO: A Stream Index for Answerng near Optmzaton Queres, full verson. Avalable at [6] 2005 UC Data Mnng Competton Homepage

Parallelism for Nested Loops with Non-uniform and Flow Dependences

Parallelism for Nested Loops with Non-uniform and Flow Dependences Parallelsm for Nested Loops wth Non-unform and Flow Dependences Sam-Jn Jeong Dept. of Informaton & Communcaton Engneerng, Cheonan Unversty, 5, Anseo-dong, Cheonan, Chungnam, 330-80, Korea. seong@cheonan.ac.kr

More information

The Greedy Method. Outline and Reading. Change Money Problem. Greedy Algorithms. Applications of the Greedy Strategy. The Greedy Method Technique

The Greedy Method. Outline and Reading. Change Money Problem. Greedy Algorithms. Applications of the Greedy Strategy. The Greedy Method Technique //00 :0 AM Outlne and Readng The Greedy Method The Greedy Method Technque (secton.) Fractonal Knapsack Problem (secton..) Task Schedulng (secton..) Mnmum Spannng Trees (secton.) Change Money Problem Greedy

More information

An Optimal Algorithm for Prufer Codes *

An Optimal Algorithm for Prufer Codes * J. Software Engneerng & Applcatons, 2009, 2: 111-115 do:10.4236/jsea.2009.22016 Publshed Onlne July 2009 (www.scrp.org/journal/jsea) An Optmal Algorthm for Prufer Codes * Xaodong Wang 1, 2, Le Wang 3,

More information

Compiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz

Compiler Design. Spring Register Allocation. Sample Exercises and Solutions. Prof. Pedro C. Diniz Compler Desgn Sprng 2014 Regster Allocaton Sample Exercses and Solutons Prof. Pedro C. Dnz USC / Informaton Scences Insttute 4676 Admralty Way, Sute 1001 Marna del Rey, Calforna 90292 pedro@s.edu Regster

More information

Learning-Based Top-N Selection Query Evaluation over Relational Databases

Learning-Based Top-N Selection Query Evaluation over Relational Databases Learnng-Based Top-N Selecton Query Evaluaton over Relatonal Databases Lang Zhu *, Wey Meng ** * School of Mathematcs and Computer Scence, Hebe Unversty, Baodng, Hebe 071002, Chna, zhu@mal.hbu.edu.cn **

More information

Virtual Memory. Background. No. 10. Virtual Memory: concept. Logical Memory Space (review) Demand Paging(1) Virtual Memory

Virtual Memory. Background. No. 10. Virtual Memory: concept. Logical Memory Space (review) Demand Paging(1) Virtual Memory Background EECS. Operatng System Fundamentals No. Vrtual Memory Prof. Hu Jang Department of Electrcal Engneerng and Computer Scence, York Unversty Memory-management methods normally requres the entre process

More information

Wishing you all a Total Quality New Year!

Wishing you all a Total Quality New Year! Total Qualty Management and Sx Sgma Post Graduate Program 214-15 Sesson 4 Vnay Kumar Kalakband Assstant Professor Operatons & Systems Area 1 Wshng you all a Total Qualty New Year! Hope you acheve Sx sgma

More information

Course Introduction. Algorithm 8/31/2017. COSC 320 Advanced Data Structures and Algorithms. COSC 320 Advanced Data Structures and Algorithms

Course Introduction. Algorithm 8/31/2017. COSC 320 Advanced Data Structures and Algorithms. COSC 320 Advanced Data Structures and Algorithms Course Introducton Course Topcs Exams, abs, Proects A quc loo at a few algorthms 1 Advanced Data Structures and Algorthms Descrpton: We are gong to dscuss algorthm complexty analyss, algorthm desgn technques

More information

Performance Evaluation of Information Retrieval Systems

Performance Evaluation of Information Retrieval Systems Why System Evaluaton? Performance Evaluaton of Informaton Retreval Systems Many sldes n ths secton are adapted from Prof. Joydeep Ghosh (UT ECE) who n turn adapted them from Prof. Dk Lee (Unv. of Scence

More information

R s s f. m y s. SPH3UW Unit 7.3 Spherical Concave Mirrors Page 1 of 12. Notes

R s s f. m y s. SPH3UW Unit 7.3 Spherical Concave Mirrors Page 1 of 12. Notes SPH3UW Unt 7.3 Sphercal Concave Mrrors Page 1 of 1 Notes Physcs Tool box Concave Mrror If the reflectng surface takes place on the nner surface of the sphercal shape so that the centre of the mrror bulges

More information

Problem Set 3 Solutions

Problem Set 3 Solutions Introducton to Algorthms October 4, 2002 Massachusetts Insttute of Technology 6046J/18410J Professors Erk Demane and Shaf Goldwasser Handout 14 Problem Set 3 Solutons (Exercses were not to be turned n,

More information

Efficient Distributed File System (EDFS)

Efficient Distributed File System (EDFS) Effcent Dstrbuted Fle System (EDFS) (Sem-Centralzed) Debessay(Debsh) Fesehaye, Rahul Malk & Klara Naherstedt Unversty of Illnos-Urbana Champagn Contents Problem Statement, Related Work, EDFS Desgn Rate

More information

The Codesign Challenge

The Codesign Challenge ECE 4530 Codesgn Challenge Fall 2007 Hardware/Software Codesgn The Codesgn Challenge Objectves In the codesgn challenge, your task s to accelerate a gven software reference mplementaton as fast as possble.

More information

6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour

6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour 6.854 Advanced Algorthms Petar Maymounkov Problem Set 11 (November 23, 2005) Wth: Benjamn Rossman, Oren Wemann, and Pouya Kheradpour Problem 1. We reduce vertex cover to MAX-SAT wth weghts, such that the

More information

Content Based Image Retrieval Using 2-D Discrete Wavelet with Texture Feature with Different Classifiers

Content Based Image Retrieval Using 2-D Discrete Wavelet with Texture Feature with Different Classifiers IOSR Journal of Electroncs and Communcaton Engneerng (IOSR-JECE) e-issn: 78-834,p- ISSN: 78-8735.Volume 9, Issue, Ver. IV (Mar - Apr. 04), PP 0-07 Content Based Image Retreval Usng -D Dscrete Wavelet wth

More information

Classifier Selection Based on Data Complexity Measures *

Classifier Selection Based on Data Complexity Measures * Classfer Selecton Based on Data Complexty Measures * Edth Hernández-Reyes, J.A. Carrasco-Ochoa, and J.Fco. Martínez-Trndad Natonal Insttute for Astrophyscs, Optcs and Electroncs, Lus Enrque Erro No.1 Sta.

More information

Feature Reduction and Selection

Feature Reduction and Selection Feature Reducton and Selecton Dr. Shuang LIANG School of Software Engneerng TongJ Unversty Fall, 2012 Today s Topcs Introducton Problems of Dmensonalty Feature Reducton Statstc methods Prncpal Components

More information

A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS

A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS Proceedngs of the Wnter Smulaton Conference M E Kuhl, N M Steger, F B Armstrong, and J A Jones, eds A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS Mark W Brantley Chun-Hung

More information

CE 221 Data Structures and Algorithms

CE 221 Data Structures and Algorithms CE 1 ata Structures and Algorthms Chapter 4: Trees BST Text: Read Wess, 4.3 Izmr Unversty of Economcs 1 The Search Tree AT Bnary Search Trees An mportant applcaton of bnary trees s n searchng. Let us assume

More information

Time- and Space-Efficient Sliding Window Top-k Query Processing

Time- and Space-Efficient Sliding Window Top-k Query Processing Tme- and Space-Effcent Sldng Wndow Top-k Query Processng KREŠIMIR PRIPUŽIĆ and IVANA PODNAR ŽARKO, Unversty of Zagreb KARL ABERER, École Polytechnque FédéraledeLausanne 1 A sldng wndow top-k (top-k/w)

More information

Sequential search. Building Java Programs Chapter 13. Sequential search. Sequential search

Sequential search. Building Java Programs Chapter 13. Sequential search. Sequential search Sequental search Buldng Java Programs Chapter 13 Searchng and Sortng sequental search: Locates a target value n an array/lst by examnng each element from start to fnsh. How many elements wll t need to

More information

Mathematics 256 a course in differential equations for engineering students

Mathematics 256 a course in differential equations for engineering students Mathematcs 56 a course n dfferental equatons for engneerng students Chapter 5. More effcent methods of numercal soluton Euler s method s qute neffcent. Because the error s essentally proportonal to the

More information

Support Vector Machines

Support Vector Machines /9/207 MIST.6060 Busness Intellgence and Data Mnng What are Support Vector Machnes? Support Vector Machnes Support Vector Machnes (SVMs) are supervsed learnng technques that analyze data and recognze patterns.

More information

Optimal Workload-based Weighted Wavelet Synopses

Optimal Workload-based Weighted Wavelet Synopses Optmal Workload-based Weghted Wavelet Synopses Yoss Matas School of Computer Scence Tel Avv Unversty Tel Avv 69978, Israel matas@tau.ac.l Danel Urel School of Computer Scence Tel Avv Unversty Tel Avv 69978,

More information

Helsinki University Of Technology, Systems Analysis Laboratory Mat Independent research projects in applied mathematics (3 cr)

Helsinki University Of Technology, Systems Analysis Laboratory Mat Independent research projects in applied mathematics (3 cr) Helsnk Unversty Of Technology, Systems Analyss Laboratory Mat-2.08 Independent research projects n appled mathematcs (3 cr) "! #$&% Antt Laukkanen 506 R ajlaukka@cc.hut.f 2 Introducton...3 2 Multattrbute

More information

An Application of the Dulmage-Mendelsohn Decomposition to Sparse Null Space Bases of Full Row Rank Matrices

An Application of the Dulmage-Mendelsohn Decomposition to Sparse Null Space Bases of Full Row Rank Matrices Internatonal Mathematcal Forum, Vol 7, 2012, no 52, 2549-2554 An Applcaton of the Dulmage-Mendelsohn Decomposton to Sparse Null Space Bases of Full Row Rank Matrces Mostafa Khorramzadeh Department of Mathematcal

More information

CSCI 104 Sorting Algorithms. Mark Redekopp David Kempe

CSCI 104 Sorting Algorithms. Mark Redekopp David Kempe CSCI 104 Sortng Algorthms Mark Redekopp Davd Kempe Algorthm Effcency SORTING 2 Sortng If we have an unordered lst, sequental search becomes our only choce If we wll perform a lot of searches t may be benefcal

More information

A Fast Content-Based Multimedia Retrieval Technique Using Compressed Data

A Fast Content-Based Multimedia Retrieval Technique Using Compressed Data A Fast Content-Based Multmeda Retreval Technque Usng Compressed Data Borko Furht and Pornvt Saksobhavvat NSF Multmeda Laboratory Florda Atlantc Unversty, Boca Raton, Florda 3343 ABSTRACT In ths paper,

More information

Steps for Computing the Dissimilarity, Entropy, Herfindahl-Hirschman and. Accessibility (Gravity with Competition) Indices

Steps for Computing the Dissimilarity, Entropy, Herfindahl-Hirschman and. Accessibility (Gravity with Competition) Indices Steps for Computng the Dssmlarty, Entropy, Herfndahl-Hrschman and Accessblty (Gravty wth Competton) Indces I. Dssmlarty Index Measurement: The followng formula can be used to measure the evenness between

More information

An Entropy-Based Approach to Integrated Information Needs Assessment

An Entropy-Based Approach to Integrated Information Needs Assessment Dstrbuton Statement A: Approved for publc release; dstrbuton s unlmted. An Entropy-Based Approach to ntegrated nformaton Needs Assessment June 8, 2004 Wllam J. Farrell Lockheed Martn Advanced Technology

More information

S1 Note. Basis functions.

S1 Note. Basis functions. S1 Note. Bass functons. Contents Types of bass functons...1 The Fourer bass...2 B-splne bass...3 Power and type I error rates wth dfferent numbers of bass functons...4 Table S1. Smulaton results of type

More information

Kent State University CS 4/ Design and Analysis of Algorithms. Dept. of Math & Computer Science LECT-16. Dynamic Programming

Kent State University CS 4/ Design and Analysis of Algorithms. Dept. of Math & Computer Science LECT-16. Dynamic Programming CS 4/560 Desgn and Analyss of Algorthms Kent State Unversty Dept. of Math & Computer Scence LECT-6 Dynamc Programmng 2 Dynamc Programmng Dynamc Programmng, lke the dvde-and-conquer method, solves problems

More information

Adaptive Load Shedding for Windowed Stream Joins

Adaptive Load Shedding for Windowed Stream Joins Adaptve Load Sheddng for Wndowed Stream Jons Bu gra Gedk College of Computng, GaTech bgedk@cc.gatech.edu Kun-Lung Wu, Phlp Yu T.J. Watson Research, IBM {klwu,psyu}@us.bm.com Lng Lu College of Computng,

More information

CHAPTER 3 SEQUENTIAL MINIMAL OPTIMIZATION TRAINED SUPPORT VECTOR CLASSIFIER FOR CANCER PREDICTION

CHAPTER 3 SEQUENTIAL MINIMAL OPTIMIZATION TRAINED SUPPORT VECTOR CLASSIFIER FOR CANCER PREDICTION 48 CHAPTER 3 SEQUENTIAL MINIMAL OPTIMIZATION TRAINED SUPPORT VECTOR CLASSIFIER FOR CANCER PREDICTION 3.1 INTRODUCTION The raw mcroarray data s bascally an mage wth dfferent colors ndcatng hybrdzaton (Xue

More information

Design and Analysis of Algorithms

Design and Analysis of Algorithms Desgn and Analyss of Algorthms Heaps and Heapsort Reference: CLRS Chapter 6 Topcs: Heaps Heapsort Prorty queue Huo Hongwe Recap and overvew The story so far... Inserton sort runnng tme of Θ(n 2 ); sorts

More information

Sorting. Sorted Original. index. index

Sorting. Sorted Original. index. index 1 Unt 16 Sortng 2 Sortng Sortng requres us to move data around wthn an array Allows users to see and organze data more effcently Behnd the scenes t allows more effectve searchng of data There are MANY

More information

Subspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points;

Subspace clustering. Clustering. Fundamental to all clustering techniques is the choice of distance measure between data points; Subspace clusterng Clusterng Fundamental to all clusterng technques s the choce of dstance measure between data ponts; D q ( ) ( ) 2 x x = x x, j k = 1 k jk Squared Eucldean dstance Assumpton: All features

More information

Problem Definitions and Evaluation Criteria for Computational Expensive Optimization

Problem Definitions and Evaluation Criteria for Computational Expensive Optimization Problem efntons and Evaluaton Crtera for Computatonal Expensve Optmzaton B. Lu 1, Q. Chen and Q. Zhang 3, J. J. Lang 4, P. N. Suganthan, B. Y. Qu 6 1 epartment of Computng, Glyndwr Unversty, UK Faclty

More information

Optimizing Document Scoring for Query Retrieval

Optimizing Document Scoring for Query Retrieval Optmzng Document Scorng for Query Retreval Brent Ellwen baellwe@cs.stanford.edu Abstract The goal of ths project was to automate the process of tunng a document query engne. Specfcally, I used machne learnng

More information

Today s Outline. Sorting: The Big Picture. Why Sort? Selection Sort: Idea. Insertion Sort: Idea. Sorting Chapter 7 in Weiss.

Today s Outline. Sorting: The Big Picture. Why Sort? Selection Sort: Idea. Insertion Sort: Idea. Sorting Chapter 7 in Weiss. Today s Outlne Sortng Chapter 7 n Wess CSE 26 Data Structures Ruth Anderson Announcements Wrtten Homework #6 due Frday 2/26 at the begnnng of lecture Proect Code due Mon March 1 by 11pm Today s Topcs:

More information

Meta-heuristics for Multidimensional Knapsack Problems

Meta-heuristics for Multidimensional Knapsack Problems 2012 4th Internatonal Conference on Computer Research and Development IPCSIT vol.39 (2012) (2012) IACSIT Press, Sngapore Meta-heurstcs for Multdmensonal Knapsack Problems Zhbao Man + Computer Scence Department,

More information

Sorting Review. Sorting. Comparison Sorting. CSE 680 Prof. Roger Crawfis. Assumptions

Sorting Review. Sorting. Comparison Sorting. CSE 680 Prof. Roger Crawfis. Assumptions Sortng Revew Introducton to Algorthms Qucksort CSE 680 Prof. Roger Crawfs Inserton Sort T(n) = Θ(n 2 ) In-place Merge Sort T(n) = Θ(n lg(n)) Not n-place Selecton Sort (from homework) T(n) = Θ(n 2 ) In-place

More information

Video Proxy System for a Large-scale VOD System (DINA)

Video Proxy System for a Large-scale VOD System (DINA) Vdeo Proxy System for a Large-scale VOD System (DINA) KWUN-CHUNG CHAN #, KWOK-WAI CHEUNG *# #Department of Informaton Engneerng *Centre of Innovaton and Technology The Chnese Unversty of Hong Kong SHATIN,

More information

High-Boost Mesh Filtering for 3-D Shape Enhancement

High-Boost Mesh Filtering for 3-D Shape Enhancement Hgh-Boost Mesh Flterng for 3-D Shape Enhancement Hrokazu Yagou Λ Alexander Belyaev y Damng We z Λ y z ; ; Shape Modelng Laboratory, Unversty of Azu, Azu-Wakamatsu 965-8580 Japan y Computer Graphcs Group,

More information

arxiv: v3 [cs.ds] 7 Feb 2017

arxiv: v3 [cs.ds] 7 Feb 2017 : A Two-stage Sketch for Data Streams Tong Yang 1, Lngtong Lu 2, Ybo Yan 1, Muhammad Shahzad 3, Yulong Shen 2 Xaomng L 1, Bn Cu 1, Gaogang Xe 4 1 Pekng Unversty, Chna. 2 Xdan Unversty, Chna. 3 North Carolna

More information

ELEC 377 Operating Systems. Week 6 Class 3

ELEC 377 Operating Systems. Week 6 Class 3 ELEC 377 Operatng Systems Week 6 Class 3 Last Class Memory Management Memory Pagng Pagng Structure ELEC 377 Operatng Systems Today Pagng Szes Vrtual Memory Concept Demand Pagng ELEC 377 Operatng Systems

More information

Chapter 6 Programmng the fnte element method Inow turn to the man subject of ths book: The mplementaton of the fnte element algorthm n computer programs. In order to make my dscusson as straghtforward

More information

Summarizing Data using Bottom-k Sketches

Summarizing Data using Bottom-k Sketches Summarzng Data usng Bottom-k Sketches Edth Cohen AT&T Labs Research 8 Park Avenue Florham Park, NJ 7932, USA edth@research.att.com Ham Kaplan School of Computer Scence Tel Avv Unversty Tel Avv, Israel

More information

Constructing Minimum Connected Dominating Set: Algorithmic approach

Constructing Minimum Connected Dominating Set: Algorithmic approach Constructng Mnmum Connected Domnatng Set: Algorthmc approach G.N. Puroht and Usha Sharma Centre for Mathematcal Scences, Banasthal Unversty, Rajasthan 304022 usha.sharma94@yahoo.com Abstract: Connected

More information

A Unified Framework for Semantics and Feature Based Relevance Feedback in Image Retrieval Systems

A Unified Framework for Semantics and Feature Based Relevance Feedback in Image Retrieval Systems A Unfed Framework for Semantcs and Feature Based Relevance Feedback n Image Retreval Systems Ye Lu *, Chunhu Hu 2, Xngquan Zhu 3*, HongJang Zhang 2, Qang Yang * School of Computng Scence Smon Fraser Unversty

More information

Tsinghua University at TAC 2009: Summarizing Multi-documents by Information Distance

Tsinghua University at TAC 2009: Summarizing Multi-documents by Information Distance Tsnghua Unversty at TAC 2009: Summarzng Mult-documents by Informaton Dstance Chong Long, Mnle Huang, Xaoyan Zhu State Key Laboratory of Intellgent Technology and Systems, Tsnghua Natonal Laboratory for

More information

Programming in Fortran 90 : 2017/2018

Programming in Fortran 90 : 2017/2018 Programmng n Fortran 90 : 2017/2018 Programmng n Fortran 90 : 2017/2018 Exercse 1 : Evaluaton of functon dependng on nput Wrte a program who evaluate the functon f (x,y) for any two user specfed values

More information

Module Management Tool in Software Development Organizations

Module Management Tool in Software Development Organizations Journal of Computer Scence (5): 8-, 7 ISSN 59-66 7 Scence Publcatons Management Tool n Software Development Organzatons Ahmad A. Al-Rababah and Mohammad A. Al-Rababah Faculty of IT, Al-Ahlyyah Amman Unversty,

More information

Query Clustering Using a Hybrid Query Similarity Measure

Query Clustering Using a Hybrid Query Similarity Measure Query clusterng usng a hybrd query smlarty measure Fu. L., Goh, D.H., & Foo, S. (2004). WSEAS Transacton on Computers, 3(3), 700-705. Query Clusterng Usng a Hybrd Query Smlarty Measure Ln Fu, Don Hoe-Lan

More information

5 The Primal-Dual Method

5 The Primal-Dual Method 5 The Prmal-Dual Method Orgnally desgned as a method for solvng lnear programs, where t reduces weghted optmzaton problems to smpler combnatoral ones, the prmal-dual method (PDM) has receved much attenton

More information

Range images. Range image registration. Examples of sampling patterns. Range images and range surfaces

Range images. Range image registration. Examples of sampling patterns. Range images and range surfaces Range mages For many structured lght scanners, the range data forms a hghly regular pattern known as a range mage. he samplng pattern s determned by the specfc scanner. Range mage regstraton 1 Examples

More information

3. CR parameters and Multi-Objective Fitness Function

3. CR parameters and Multi-Objective Fitness Function 3 CR parameters and Mult-objectve Ftness Functon 41 3. CR parameters and Mult-Objectve Ftness Functon 3.1. Introducton Cogntve rados dynamcally confgure the wreless communcaton system, whch takes beneft

More information

A Deflected Grid-based Algorithm for Clustering Analysis

A Deflected Grid-based Algorithm for Clustering Analysis A Deflected Grd-based Algorthm for Clusterng Analyss NANCY P. LIN, CHUNG-I CHANG, HAO-EN CHUEH, HUNG-JEN CHEN, WEI-HUA HAO Department of Computer Scence and Informaton Engneerng Tamkang Unversty 5 Yng-chuan

More information

Some material adapted from Mohamed Younis, UMBC CMSC 611 Spr 2003 course slides Some material adapted from Hennessy & Patterson / 2003 Elsevier

Some material adapted from Mohamed Younis, UMBC CMSC 611 Spr 2003 course slides Some material adapted from Hennessy & Patterson / 2003 Elsevier Some materal adapted from Mohamed Youns, UMBC CMSC 611 Spr 2003 course sldes Some materal adapted from Hennessy & Patterson / 2003 Elsever Scence Performance = 1 Executon tme Speedup = Performance (B)

More information

TPL-Aware Displacement-driven Detailed Placement Refinement with Coloring Constraints

TPL-Aware Displacement-driven Detailed Placement Refinement with Coloring Constraints TPL-ware Dsplacement-drven Detaled Placement Refnement wth Colorng Constrants Tao Ln Iowa State Unversty tln@astate.edu Chrs Chu Iowa State Unversty cnchu@astate.edu BSTRCT To mnmze the effect of process

More information

Summarization and Matching of Density-Based Clusters in Streaming Environments

Summarization and Matching of Density-Based Clusters in Streaming Environments Summarzaton and Matchng of Densty-Based Clusters n Streamng Envronments D Yang Oracle Corporaton 1 Oracle Drve Nashua, NH, USA d.yang@oracle.com Elke A. Rundenstener Worcester Polytechnc Insttute 100 Insttute

More information

Support Vector Machines

Support Vector Machines Support Vector Machnes Decson surface s a hyperplane (lne n 2D) n feature space (smlar to the Perceptron) Arguably, the most mportant recent dscovery n machne learnng In a nutshell: map the data to a predetermned

More information

CPU Load Shedding for Binary Stream Joins

CPU Load Shedding for Binary Stream Joins Under consderaton for publcaton n Knowledge and Informaton Systems CPU Load Sheddng for Bnary Stream Jons Bugra Gedk 1,2, Kun-Lung Wu 1, Phlp S. Yu 1 and Lng Lu 2 1 IBM T.J. Watson Research Center, Hawthorne,

More information

Solving two-person zero-sum game by Matlab

Solving two-person zero-sum game by Matlab Appled Mechancs and Materals Onlne: 2011-02-02 ISSN: 1662-7482, Vols. 50-51, pp 262-265 do:10.4028/www.scentfc.net/amm.50-51.262 2011 Trans Tech Publcatons, Swtzerland Solvng two-person zero-sum game by

More information

Machine Learning: Algorithms and Applications

Machine Learning: Algorithms and Applications 14/05/1 Machne Learnng: Algorthms and Applcatons Florano Zn Free Unversty of Bozen-Bolzano Faculty of Computer Scence Academc Year 011-01 Lecture 10: 14 May 01 Unsupervsed Learnng cont Sldes courtesy of

More information

Run-Time Operator State Spilling for Memory Intensive Long-Running Queries

Run-Time Operator State Spilling for Memory Intensive Long-Running Queries Run-Tme Operator State Spllng for Memory Intensve Long-Runnng Queres Bn Lu, Yal Zhu, and lke A. Rundenstener epartment of Computer Scence, Worcester Polytechnc Insttute Worcester, Massachusetts, USA {bnlu,

More information

A mathematical programming approach to the analysis, design and scheduling of offshore oilfields

A mathematical programming approach to the analysis, design and scheduling of offshore oilfields 17 th European Symposum on Computer Aded Process Engneerng ESCAPE17 V. Plesu and P.S. Agach (Edtors) 2007 Elsever B.V. All rghts reserved. 1 A mathematcal programmng approach to the analyss, desgn and

More information

A Binarization Algorithm specialized on Document Images and Photos

A Binarization Algorithm specialized on Document Images and Photos A Bnarzaton Algorthm specalzed on Document mages and Photos Ergna Kavalleratou Dept. of nformaton and Communcaton Systems Engneerng Unversty of the Aegean kavalleratou@aegean.gr Abstract n ths paper, a

More information

Adaptive Load Shedding for Windowed Stream Joins

Adaptive Load Shedding for Windowed Stream Joins Adaptve Load Sheddng for Wndowed Stream Jons Buğra Gedk, Kun-Lung Wu, Phlp S. Yu, Lng Lu College of Computng, Georga Tech Atlanta GA 333 {bgedk,lnglu}@cc.gatech.edu IBM T. J. Watson Research Center Yorktown

More information

Smoothing Spline ANOVA for variable screening

Smoothing Spline ANOVA for variable screening Smoothng Splne ANOVA for varable screenng a useful tool for metamodels tranng and mult-objectve optmzaton L. Rcco, E. Rgon, A. Turco Outlne RSM Introducton Possble couplng Test case MOO MOO wth Game Theory

More information

Learning the Kernel Parameters in Kernel Minimum Distance Classifier

Learning the Kernel Parameters in Kernel Minimum Distance Classifier Learnng the Kernel Parameters n Kernel Mnmum Dstance Classfer Daoqang Zhang 1,, Songcan Chen and Zh-Hua Zhou 1* 1 Natonal Laboratory for Novel Software Technology Nanjng Unversty, Nanjng 193, Chna Department

More information

CS 534: Computer Vision Model Fitting

CS 534: Computer Vision Model Fitting CS 534: Computer Vson Model Fttng Sprng 004 Ahmed Elgammal Dept of Computer Scence CS 534 Model Fttng - 1 Outlnes Model fttng s mportant Least-squares fttng Maxmum lkelhood estmaton MAP estmaton Robust

More information

Assignment # 2. Farrukh Jabeen Algorithms 510 Assignment #2 Due Date: June 15, 2009.

Assignment # 2. Farrukh Jabeen Algorithms 510 Assignment #2 Due Date: June 15, 2009. Farrukh Jabeen Algorthms 51 Assgnment #2 Due Date: June 15, 29. Assgnment # 2 Chapter 3 Dscrete Fourer Transforms Implement the FFT for the DFT. Descrbed n sectons 3.1 and 3.2. Delverables: 1. Concse descrpton

More information

Simulation Based Analysis of FAST TCP using OMNET++

Simulation Based Analysis of FAST TCP using OMNET++ Smulaton Based Analyss of FAST TCP usng OMNET++ Umar ul Hassan 04030038@lums.edu.pk Md Term Report CS678 Topcs n Internet Research Sprng, 2006 Introducton Internet traffc s doublng roughly every 3 months

More information

Motivation. EE 457 Unit 4. Throughput vs. Latency. Performance Depends on View Point?! Computer System Performance. An individual user wants to:

Motivation. EE 457 Unit 4. Throughput vs. Latency. Performance Depends on View Point?! Computer System Performance. An individual user wants to: 4.1 4.2 Motvaton EE 457 Unt 4 Computer System Performance An ndvdual user wants to: Mnmze sngle program executon tme A datacenter owner wants to: Maxmze number of Mnmze ( ) http://e-tellgentnternetmarketng.com/webste/frustrated-computer-user-2/

More information

Outline. Discriminative classifiers for image recognition. Where in the World? A nearest neighbor recognition example 4/14/2011. CS 376 Lecture 22 1

Outline. Discriminative classifiers for image recognition. Where in the World? A nearest neighbor recognition example 4/14/2011. CS 376 Lecture 22 1 4/14/011 Outlne Dscrmnatve classfers for mage recognton Wednesday, Aprl 13 Krsten Grauman UT-Austn Last tme: wndow-based generc obect detecton basc ppelne face detecton wth boostng as case study Today:

More information

CS246: Mining Massive Datasets Jure Leskovec, Stanford University

CS246: Mining Massive Datasets Jure Leskovec, Stanford University CS46: Mnng Massve Datasets Jure Leskovec, Stanford Unversty http://cs46.stanford.edu /19/013 Jure Leskovec, Stanford CS46: Mnng Massve Datasets, http://cs46.stanford.edu Perceptron: y = sgn( x Ho to fnd

More information

Cache Performance 3/28/17. Agenda. Cache Abstraction and Metrics. Direct-Mapped Cache: Placement and Access

Cache Performance 3/28/17. Agenda. Cache Abstraction and Metrics. Direct-Mapped Cache: Placement and Access Agenda Cache Performance Samra Khan March 28, 217 Revew from last lecture Cache access Assocatvty Replacement Cache Performance Cache Abstracton and Metrcs Address Tag Store (s the address n the cache?

More information

Lecture 5: Multilayer Perceptrons

Lecture 5: Multilayer Perceptrons Lecture 5: Multlayer Perceptrons Roger Grosse 1 Introducton So far, we ve only talked about lnear models: lnear regresson and lnear bnary classfers. We noted that there are functons that can t be represented

More information

NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS

NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS ARPN Journal of Engneerng and Appled Scences 006-017 Asan Research Publshng Network (ARPN). All rghts reserved. NUMERICAL SOLVING OPTIMAL CONTROL PROBLEMS BY THE METHOD OF VARIATIONS Igor Grgoryev, Svetlana

More information

CMPS 10 Introduction to Computer Science Lecture Notes

CMPS 10 Introduction to Computer Science Lecture Notes CPS 0 Introducton to Computer Scence Lecture Notes Chapter : Algorthm Desgn How should we present algorthms? Natural languages lke Englsh, Spansh, or French whch are rch n nterpretaton and meanng are not

More information

FORESIGHTED JOINT RESOURCE RECIPROCATION AND SCHEDULING STRATEGIES FOR REAL-TIME VIDEO STREAMING OVER PEER-TO-PEER NETWORKS

FORESIGHTED JOINT RESOURCE RECIPROCATION AND SCHEDULING STRATEGIES FOR REAL-TIME VIDEO STREAMING OVER PEER-TO-PEER NETWORKS FORESIGHTED JOINT RESOURCE RECIPROCATION AND SCHEDULING STRATEGIES FOR REAL-TIME VIDEO STREAMING OVER PEER-TO-PEER NETWORKS Sunghoon Ivan Lee, Hyunggon Park, and Mhaela van der Schaar Electrcal Engneerng

More information

Sum of Linear and Fractional Multiobjective Programming Problem under Fuzzy Rules Constraints

Sum of Linear and Fractional Multiobjective Programming Problem under Fuzzy Rules Constraints Australan Journal of Basc and Appled Scences, 2(4): 1204-1208, 2008 ISSN 1991-8178 Sum of Lnear and Fractonal Multobjectve Programmng Problem under Fuzzy Rules Constrants 1 2 Sanjay Jan and Kalash Lachhwan

More information

Analysis of Collaborative Distributed Admission Control in x Networks

Analysis of Collaborative Distributed Admission Control in x Networks 1 Analyss of Collaboratve Dstrbuted Admsson Control n 82.11x Networks Thnh Nguyen, Member, IEEE, Ken Nguyen, Member, IEEE, Lnha He, Member, IEEE, Abstract Wth the recent surge of wreless home networks,

More information

Cluster Analysis of Electrical Behavior

Cluster Analysis of Electrical Behavior Journal of Computer and Communcatons, 205, 3, 88-93 Publshed Onlne May 205 n ScRes. http://www.scrp.org/ournal/cc http://dx.do.org/0.4236/cc.205.350 Cluster Analyss of Electrcal Behavor Ln Lu Ln Lu, School

More information

Load-Balanced Anycast Routing

Load-Balanced Anycast Routing Load-Balanced Anycast Routng Chng-Yu Ln, Jung-Hua Lo, and Sy-Yen Kuo Department of Electrcal Engneerng atonal Tawan Unversty, Tape, Tawan sykuo@cc.ee.ntu.edu.tw Abstract For fault-tolerance and load-balance

More information

Private Information Retrieval (PIR)

Private Information Retrieval (PIR) 2 Levente Buttyán Problem formulaton Alce wants to obtan nformaton from a database, but she does not want the database to learn whch nformaton she wanted e.g., Alce s an nvestor queryng a stock-market

More information

3D vector computer graphics

3D vector computer graphics 3D vector computer graphcs Paolo Varagnolo: freelance engneer Padova Aprl 2016 Prvate Practce ----------------------------------- 1. Introducton Vector 3D model representaton n computer graphcs requres

More information

User Authentication Based On Behavioral Mouse Dynamics Biometrics

User Authentication Based On Behavioral Mouse Dynamics Biometrics User Authentcaton Based On Behavoral Mouse Dynamcs Bometrcs Chee-Hyung Yoon Danel Donghyun Km Department of Computer Scence Department of Computer Scence Stanford Unversty Stanford Unversty Stanford, CA

More information

Quality Improvement Algorithm for Tetrahedral Mesh Based on Optimal Delaunay Triangulation

Quality Improvement Algorithm for Tetrahedral Mesh Based on Optimal Delaunay Triangulation Intellgent Informaton Management, 013, 5, 191-195 Publshed Onlne November 013 (http://www.scrp.org/journal/m) http://dx.do.org/10.36/m.013.5601 Qualty Improvement Algorthm for Tetrahedral Mesh Based on

More information

Random Kernel Perceptron on ATTiny2313 Microcontroller

Random Kernel Perceptron on ATTiny2313 Microcontroller Random Kernel Perceptron on ATTny233 Mcrocontroller Nemanja Djurc Department of Computer and Informaton Scences, Temple Unversty Phladelpha, PA 922, USA nemanja.djurc@temple.edu Slobodan Vucetc Department

More information

Intra-Parametric Analysis of a Fuzzy MOLP

Intra-Parametric Analysis of a Fuzzy MOLP Intra-Parametrc Analyss of a Fuzzy MOLP a MIAO-LING WANG a Department of Industral Engneerng and Management a Mnghsn Insttute of Technology and Hsnchu Tawan, ROC b HSIAO-FAN WANG b Insttute of Industral

More information

A Robust Method for Estimating the Fundamental Matrix

A Robust Method for Estimating the Fundamental Matrix Proc. VIIth Dgtal Image Computng: Technques and Applcatons, Sun C., Talbot H., Ourseln S. and Adraansen T. (Eds.), 0- Dec. 003, Sydney A Robust Method for Estmatng the Fundamental Matrx C.L. Feng and Y.S.

More information

The Research of Support Vector Machine in Agricultural Data Classification

The Research of Support Vector Machine in Agricultural Data Classification The Research of Support Vector Machne n Agrcultural Data Classfcaton Le Sh, Qguo Duan, Xnmng Ma, Me Weng College of Informaton and Management Scence, HeNan Agrcultural Unversty, Zhengzhou 45000 Chna Zhengzhou

More information

CSE 326: Data Structures Quicksort Comparison Sorting Bound

CSE 326: Data Structures Quicksort Comparison Sorting Bound CSE 326: Data Structures Qucksort Comparson Sortng Bound Steve Setz Wnter 2009 Qucksort Qucksort uses a dvde and conquer strategy, but does not requre the O(N) extra space that MergeSort does. Here s the

More information

GSLM Operations Research II Fall 13/14

GSLM Operations Research II Fall 13/14 GSLM 58 Operatons Research II Fall /4 6. Separable Programmng Consder a general NLP mn f(x) s.t. g j (x) b j j =. m. Defnton 6.. The NLP s a separable program f ts objectve functon and all constrants are

More information

Simplification of 3D Meshes

Simplification of 3D Meshes Smplfcaton of 3D Meshes Addy Ngan /4/00 Outlne Motvaton Taxonomy of smplfcaton methods Hoppe et al, Mesh optmzaton Hoppe, Progressve meshes Smplfcaton of 3D Meshes 1 Motvaton Hgh detaled meshes becomng

More information

Computer Animation and Visualisation. Lecture 4. Rigging / Skinning

Computer Animation and Visualisation. Lecture 4. Rigging / Skinning Computer Anmaton and Vsualsaton Lecture 4. Rggng / Sknnng Taku Komura Overvew Sknnng / Rggng Background knowledge Lnear Blendng How to decde weghts? Example-based Method Anatomcal models Sknnng Assume

More information

Outline. Self-Organizing Maps (SOM) US Hebbian Learning, Cntd. The learning rule is Hebbian like:

Outline. Self-Organizing Maps (SOM) US Hebbian Learning, Cntd. The learning rule is Hebbian like: Self-Organzng Maps (SOM) Turgay İBRİKÇİ, PhD. Outlne Introducton Structures of SOM SOM Archtecture Neghborhoods SOM Algorthm Examples Summary 1 2 Unsupervsed Hebban Learnng US Hebban Learnng, Cntd 3 A

More information

CSE 326: Data Structures Quicksort Comparison Sorting Bound

CSE 326: Data Structures Quicksort Comparison Sorting Bound CSE 326: Data Structures Qucksort Comparson Sortng Bound Bran Curless Sprng 2008 Announcements (5/14/08) Homework due at begnnng of class on Frday. Secton tomorrow: Graded homeworks returned More dscusson

More information