Maintaining Multi-Modality through Mixture Tracking
|
|
- Shannon Harvey
- 8 years ago
- Views:
Transcription
1 Mainaining Muli-Modaliy hrough Mixure Tracking Jaco Vermaak, Arnaud Douce Cambridge Universiy Engineering Deparmen Cambridge, CB2 1PZ, UK Parick Pérez Microsof Research Cambridge, CB3 0FB, UK Absrac In recen years paricle filers have become a remendously popular ool o perform racking for non-linear and/or non-gaussian models. This is due o heir simpliciy, generaliy and success over a wide range of challenging applicaions. Paricle filers, and Mone Carlo mehods in general, are however poor a consisenly mainaining he muli-modaliy of he arge disribuions ha may arise due o ambiguiy or he presence of muliple objecs. To address his shorcoming his paper proposes o model he arge disribuion as a non-parameric mixure model, and presens he general racking recursion in his case. I is shown how a Mone Carlo implemenaion of he general recursion leads o a mixure of paricle filers ha inerac only in he compuaion of he mixure weighs, hus leading o an efficien numerical algorihm, where all he resuls peraining o sandard paricle filers apply. The abiliy of he new mehod o mainain poserior muli-modaliy is illusraed on a synheic example and a real world racking problem involving he racking of fooball players in a video sequence. 1. Inroducion Tracking involves he deecion and recursive localisaion of an objec or objecs of ineres based on sequenial daa measuremens. Typical examples include face racking in video sequences [6], racking aircraf using radar reurns [2], localising a mobile robo using laser range measuremens [12], and many more. In pracical seings here are many facors ha conribue owards he uncerainy in an objec s exac locaion and configuraion. These include measuremen noise, inaccurae modelling, cluer (false posiives), flucuaions in environmenal condiions, ec. To adequaely capure he uncerainy due o hese facors a probabilisic framework is required. Wihin a racking conex one paricularly popular approach is Bayesian Sequenial Esimaion. This framework allows he recursive esimaion of a ime-evolving poserior disribuion ha describes he objec sae condiional on all he observaions seen so far, commonly known as he filering disribuion. I requires he definiion of a Markovian dynamic model ha describes how he objec sae evolves, and a model o evaluae he likelihood of a hypohesised sae giving rise o he observed daa. This, in heory, is sufficien o allow recursive esimaion of he filering disribuion. However, he likelihood models for racking ofen lead o inracable inference, requiring approximaion echniques. In recen years Sequenial Mone Carlo Esimaion, oherwise known as Paricle Filering [4], has proved o be a popular approximaion mehodology. Is populariy sems from is simpliciy, generaliy and success over a wide range of challenging applicaions. I represens he filering disribuion wih a se of samples, or paricles, and associaed imporance weighs, which are hen propagaed hrough ime o give approximaions of he filering disribuion a subsequen ime seps. I requires only he definiion of a suiable proposal disribuion from which new paricles can be simulaed, and he abiliy o evaluae he likelihood and dynamic models. One imporan shorcoming of paricle filers, and Mone Carlo mehods in general, is ha hey are poor a consisenly mainaining he muli-modaliy in he arge disribuion. Muliple modes arise if here is ambiguiy abou he objec sae due o insufficien measuremens or cluer, or if he measuremens come from muliple objecs. In he firs case i is desirable o rack all he modes unil he ambiguiy can be naurally resolved, and in he second, i is ofen required o rack all he objecs presen. In a pracical paricle filer implemenaion, however, i ofen happens ha all he paricles quickly migrae o one of he modes, subsequenly discarding all oher modes. This paper inroduces a sraegy ha is beer able o mainain muli-modaliy. Working from he assumpion ha mixure models are inherenly more effecive a capuring muliple modes, he arge disribuion is formulaed as a non-parameric mixure of filering disribuions. As is he case for Bayesian sequenial esimaion a general framework is derived in which he mixure filering disribuion Proceedings of he Ninh IEEE Inernaional Conference on Compuer Vision (ICCV 2003) 2-Volume Se
2 can be compued recursively in wo seps: a predicion sep, followed by an updae sep when he new daa becomes available. A Mone Carlo implemenaion of he general framework essenially leads o a mixure of paricle filers ha inerac only in he compuaion of he mixure weighs. This resul is elegan in he sense ha all he resuls ha hold for sandard paricle filers ransfer o he individual mixure componens. Furhermore, he approach can be combined wih any convenien sraegy o obain and updae he mixure represenaion. The remainder of he paper is organised as follows. Secion 2 inroduces he general mixure racking framework. Secion 3 shows how a Mone Carlo implemenaion of he general framework leads o a mixure of paricle filers. Secion 4 discusses some imporan issues regarding he iniialisaion and updaing of he mixure represenaion. In Secion 5 he proposed algorihm is compared o he sandard paricle filer on wo problems. The firs is a synheic example where he muli-modaliy is due o ambiguiy, whereas he second is a real world problem involving he racking of muliple fooball players in a video sequence. The paper is summarised in Secion 6. Relaed Work The weakness of paricle mehods o mainain mulimodaliy has been acknowledged before. The auhors in [9] inroduce he idea of clusered paricle filering o guard agains samples ses becoming premaurely impoverished in he conex of mobile robo localisaion in highly symmeric environmens. The algorihm essenially groups he paricles ino clusers ha are independenly racked. Each cluser is assigned a probabiliy ha is racked using anoher Mone Carlo filer operaing a a higher level. The cluser wih he highes probabiliy a any paricular ime sep is deemed o correspond o he rue robo configuraion a ha insan. In he special conex of muli-objec racking a vas body of lieraure exiss. However, mos algorihms broadly fall ino one of wo caegories. The firs builds muli-objec rackers by muliple insaniaions of single objec racking algorihms, e.g. [3, 11]. Sraegies wih various levels of sophisicaion have been developed o inerpre he oupu of he resuling rackers in he case of occlusions and overlapping objecs. The second caegory of muli-objec rackers explicily exends he sae-space o include componens for all he objecs of ineres, e.g. [7, 8]. A variable number of objecs can be accommodaed by eiher dynamically changing he dimension of he sae-space, or by a corresponding se of indicaor variables signifying wheher an objec is presen or no. 2. Mixure Tracking Le x denoe he sae of he objec of ineres, and y = (y 1 y ) he observaions up o ime. For racking he disribuion of ineres is he filering disribuion p(x y ). In Bayesian sequenial esimaion his disribuion can be compued using he wo sep recursion: predic: p(x y 1 )= D(x x 1 )p(dx 1 y 1 ) (1) updae: p(x y )= L(y x )p(x y 1 ) L(y s )p(ds y 1 ), (2) where he predicion disribuion follows from marginalisaion, and he new filering disribuion is a direc consequence of Bayes rule. The recursion requires he specificaion of a dynamic model describing he sae evoluion D(x x 1 ), and a model ha gives he likelihood of any sae in he ligh of he curren observaion L(y x ). The recursion is iniialised wih some iniial disribuion p(x 0 ). To capure muli-modaliy his paper formulaes he filering disribuion as a M-componen mixure model, i.e. p(x y )= M π m, p m (x y ), (3) wih M π m, =1. Noe ha no parameric model is assumed for he individual mixure componens. The remainder of his secion shows how his non-parameric mixure represenaion can be updaed recursively in he same fashion as he wo sep approach for sandard Bayesian sequenial esimaion. Assuming ha he mixure filering disribuion p(x 1 y 1 ) is known, he new predicion disribuion is obained by subsiuion ino (1), leading o p(x y 1 )= = M π m, 1 D(x x 1 )p m (dx 1 y 1 ) M π m, 1 p m (x y 1 ), wih p m (x y 1 ) = D(x x 1 )p m (dx 1 y 1 ) he predicion disribuion for he m-h componen. Thus he new predicion disribuion is sraighforwardly obained by compuing he predicion disribuion for each of he componens individually, and combining hem in a mixure ha reains he original componen weighs. To obain he new filering disribuion he predicion dis- Proceedings of he Ninh IEEE Inernaional Conference on Compuer Vision (ICCV 2003) 2-Volume Se
3 ribuion is subsiued ino (2), leading o M p(x y )= π m, 1L(y x )p m (x y 1 ) M n=1 π n, 1 L(y s )p n (ds y 1 ) M [ π m, 1 L(y s )p m (ds y 1 ) ] = M n=1 π n, 1 L(y s )p n (ds y 1 ) [ L(y x )p m (x y 1 ) ]. L(y s )p m (ds y 1 ) In he second line he second erm in brackes is easily recognised as he new filering disribuion for he m-h componen, i.e. p m (x y )= L(y x )p m (x y 1 ) L(y s )p m (ds y 1 ). The firs erm in brackes is independen of he sae x, and if his is aken o be he new weigh, i.e. π m, 1 L(y s )p m (ds y 1 ) π m, = M n=1 π n, 1 L(y s )p n (ds y 1 ) (4) π m, 1 p m (y y 1 ) = M n=1 π n, 1p n (y y 1 ), he new filering disribuion is again a mixure of he individual componen filering disribuions, as in (3). This is an elegan resul indeed and means ha he filering recursion can be performed for each componen individually. The correc arge disribuion is mainained as long as he mixure weighs are updaed according o (4). The new componen weigh is he normalised weighed likelihood for he componen. This is he only par of he procedure where he componens inerac. 3. Paricle Approximaion The general mixure racking recursion inroduced in he previous secion yields closed-form expressions in only a small number of cases. A noable example occurs if boh he dynamic and likelihood models are linear and Gaussian, resuling in a mixure of Kalman filers [1]. For models ha are non-linear and/or non-gaussian approximaion echniques are required. One popular approximaion sraegy is Sequenial Mone Carlo mehods [4], oherwise known as Paricle Filers. These have gained remendous populariy in recen years as a numerical approximaion sraegy for complex models. This is due o heir simpliciy, generaliy and modelling success over a wide range of challenging applicaions. The paricle filer is a Mone Carlo mehod ha represens he arge disribuion wih a weighed se of samples ha are propagaed in such a manner so as o mainain a properly weighed sample from he arge disribuion a subsequen ime seps. The remainder of his secion presens he deails of a paricle approximaion o he general mixure racking recursion. In wha follows le P = {N,M,Π, X, W, C } denoe he paricle represenaion of he mixure filering disribuion in (3), wih N he number of paricles, M he number of mixure componens, Π = {π m, } M he mixure componen weighs, X = {x (i) { } N i=1 he paricle weighs, and C = {c (i) } N i=1 he paricles, W = } N i=1 he com- {1 M}, wih c (i) = m if ponen indicaors, i.e. c (i) paricle i belongs o mixure componen m. The paricle represenaion implies a Mone Carlo approximaion of he mixure filering disribuion of he form p(x y )= M π m, δ (i) x (x ), where δ a ( ) is he Dirac dela measure wih mass a a, and I m = {i {1 N} : c (i) = m} is he se of indices of he paricles belonging o he m-h mixure componen. Noe ha he mixure componen weighs and he paricle weighs for each mixure componen sum o one, i.e. M π m, = 1 and =1, m =1 M. Given a paricle se P 1 ha is approximaely disribued according o p(x 1 y 1 ), he objecive is o compue he new paricle se P such ha i is a sample se from p(x y ). Recall ha in he general mixure racking recursion each mixure componen evolves independenly, and ha he mixure componens inerac only in he compuaion of he mixure weighs. In he same way he paricle represenaions for he mixure componens also evolve independenly. Considering he m-h componen, he samples {x (i) 1,w(i) 1 } is a properly weighed sample se from p m (x 1 y 1 ). New samples are generaed from a suiably chosen proposal disribuion, which may depend on he old sae and he new measuremen, i.e. x (i) q(x x (i) 1, y ), i I m. To mainain a properly weighed sample se he new paricle weighs are se o = j I m w (j), = w(i) 1 L(y x (i) )D(x (i) x (i) q(x (i) x (i) 1, y ) The new sample se {x (i), } i Im is hen approximaely disribued o p m (x y ). To obain he new mixure weighs i is necessary o compue he componen likelihoods p m (y y 1 ), m = 1 M. Using he paricles a Mone Carlo approximaion 1 ). Proceedings of he Ninh IEEE Inernaional Conference on Compuer Vision (ICCV 2003) 2-Volume Se
4 o he m-h componen likelihood can be obained as p m (y y 1 )= L(y x )D(dx x 1 )p m (dx 1 y 1 ) 1 L(y x (i) )D(x (i) q(x (i) x (i) 1, y ) x (i) 1 ) =. Subsiuing his resul ino (4) leads o an approximaion for he new mixure weighs given by π m, π m, 1 w m, M n=1 π n, 1 w n,, w m, =. From ime o ime i is necessary o resample he paricles o avoid degeneracy of he weighs (see [5] for more deails on degeneracy and resampling procedures). Sandard resampling, however, is insensiive o he locaions of he paricles, and may lead o a loss of suppor in he arge disribuion. A very imporan poin o noe here is ha he mixure modelling approach allows independen resampling of each of he mixure componens according o he componen paricle weighs, hus naurally preserving he suppor of he poserior disribuion. Following such a procedure he new paricle weighs become =1/ I m, i I m, where denoes he se size operaor. 4. Mixure Compuaion The discussion up o his poin showed how a mixure represenaion for he filering disribuion can be propagaed, once such a represenaion is available. So far nohing has been said abou how o obain and mainain he mixure represenaion. In he ideal case here would be one mixure componen for each of he modes in he arge disribuion. In pracice, however, he number of modes in he arge disribuion is rarely known beforehand. Furhermore, he number of modes is unlikely o remain fixed, bu may flucuae as ambiguiies arise and are resolved, or objecs appear and disappear. Thus, from ime o ime i is necessary o recompue he mixure represenaion o ake accoun of hese flucuaions. For example, i may be desirable o merge componens ha have a significan degree of overlap, and spli componens ha have become oo diffuse. Forunaely his is easy o achieve using he paricle represenaion. Denoe by (C,M ) = F(X, C,M) a spaial reclusering procedure. I akes as inpus he paricles and he curren mixure represenaion (componen indicaors and number of componens), and compues a new mixure represenaion ha may or may no have he same number of componens as he original represenaion. Such a funcion encapsulaes any mixure compuaion operaion of ineres, including merging, spliing, reclusering ec. Wha remains is o compue he new mixure and paricle weighs, Π and W, so ha he new mixure approximaion P = {N,M, Π, X, W, C } is equal in disribuion o P. These are sraighforwardly obained by developing he mixure represenaion for P as follows: p(x y )= = = N i=1 M π c (i) M π m,, w(i) π m, δ (i) x (x )= w (i) δ (i) x δ (i) x (x ) M (x ), π c (i), w(i) where he new mixure and paricle weighs are given by π m, = i I m π (i) c, w(i), w (i) = π c (i), w(i) π c (i), δ (i) x (x ) Wih hese weighs he recompued mixure P represens exacly he same disribuion as P, and can be subsiued for P wihou affecing he convergence properies of he paricle filer. Noe ha he paricles X are no affeced in he new represenaion. The reclusering funcion F can be implemened in any convenien way. For he applicaions in his paper he iniial mixure represenaion is obained by K-means clusering of he iniial sample se, which is simulaed from some iniial sae disribuion p(x 0 ). A each ieraion he mixure represenaion is recompued by merging clusers wih significan overlap, and spliing clusers ha have become oo diffuse. Following his he new mixure componens are refined by a run of he K-means algorihm, iniialised wih he componens obained afer he merging and spliing procedures. This simple sraegy ha allows boh he mixure composiion and he number of mixure componens o vary was found o work well in he applicaions considered here, as illusraed in he following secion. 5. Experimens and Resuls This secion compares he performance of he proposed mixure paricle filer wih ha of he sandard paricle filer on wo muli-modal racking problems. The firs is a synheic example where he mulimodaliy is due o ambiguiy. The second is a real world problem involving he racking of muliple fooball players in a video sequence Synheic Example The purpose of his example is o esablish a baseline performance comparison beween he sandard and mixure paricle filers on a problem where he ground ruh is. Proceedings of he Ninh IEEE Inernaional Conference on Compuer Vision (ICCV 2003) 2-Volume Se
5 known. The model considered is scalar, and he governing equaions are given by D(x x 1 )=N(x x 1,σx) 2 L(y x )=N(y x 2,σy), 2 where N( µ, σ 2 ) denoes he univariae Gaussian disribuion wih mean µ and variance σ 2. In he resuls repored here he parameers were se o σ x = σ y = 0.1. The symmery of he Gaussian random walk dynamics and he quadraic erm in he likelihood means ha he filering disribuion has wo modes of equal mass 1 a ±x, wih x he rue sae. An excepion occurs when x is close o zero, in which case he wo modes merge, resuling in a single mode a zero. Figure 1 shows some synheic daa for 100 ime seps. Noe ha he rue sae was no simulaed from he model, bu deerminisically generaed o be sinusoidal. The abiliy of he sandard and mixure paricle filers o mainain he ambiguiy was esed on his daa se. In boh cases he iniial paricles were uniformly generaed, and he paricle proposal was aken o be he dynamics, so ha he new paricle weighs become proporional o he old weighs muliplied by he corresponding likelihoods. The mixure paricle filer was consrained o have a maximum of wo componens. The comparaive resuls for a ypical run wih 100 paricles are given in Figure 1. The sandard paricle filer loses rack of he rue mode afer ime sep 60. This behaviour is common in paricle filers, and Mone Carlo mehods in general. The mixure paricle filer, however, is able o successfully rack boh modes hroughou. To deermine he generaliy of his resul he experimen was repeaed 20 imes for differen numbers of paricles. For each run he performance score was defined as he fracion of ime seps during which boh modes were represened, hus ranging from zero (only one mode racked hroughou) o one (boh modes racked hroughou). The resuls are given in Figure 2. As expeced he performance of he sandard paricle filer increases wih an increase in he number of paricles, up o 500 paricles, from which poin i is able o consisenly rack boh modes. Even wih a small number of paricles he mixure paricle filer never fails o rack boh modes. I is ineresing o noe he behaviour of he firs mixure componen weigh, also depiced in Figure 2. Since boh modes are equally srong he mean weigh is approximaely 0.5 over all he ime seps, regardless of he number of paricles. The variabiliy in he weigh behaviour, however, decreases wih an increase in he number of paricles. 1 Assuming ha he iniial sae disribuion is also symmeric ime sep rue sae false sae observaion ime sep Figure 1. Simulaed paricles. (Top) Sandard paricle filer wih 100 paricles. (Boom) Mixure paricle filer wih a maximum of wo mixure componens and 50 paricles per componen. The sandard paricle filer loses rack of he rue mode afer ime sep 60, whereas he mixure paricle filer racks boh modes hroughou Visual Tracking This secion considers he problem of racking fooball players in a video sequence. In his seing he mulimodaliy is due o he presence of muliple objecs (fooball players), and o some exen, cluer. More precisely, he objec of ineres is represened by is bounding box. However, more general objec models can easily be accommodaed. The reference bounding box o be racked is specified by he user, and parameerised as 2 B ref =(x ref,y ref,l x,l y ), where (x ref,y ref ) is he cenre of he bounding box, and l x and l y are he bounding box widh and heigh, respecively. For he racking he sae of he bounding box is aken o be x =(x, y, s x,s y ),so 2 In wha follows he ime subscrip is suppressed for he sake of breviy. Proceedings of he Ninh IEEE Inernaional Conference on Compuer Vision (ICCV 2003) 2-Volume Se
6 1 0.8 likelihood for a hypohesised sae is defined as [ L(y x,b ref ) exp ( B 2 (h R B x, h R B ref ) + B 2 (h G B x, h G B ref )+B 2 (h B B x, h B B ref ) ) /2σ 2], weigh score paricles PF MPF ime sep Figure 2. Muliple runs. (Top) Score curves. Performance score and error bars as a funcion of he number of paricles. (Boom) Mixure componen one weigh behaviour for an increasing number of paricles. The mean weigh is indicaed by a solid line, and is approximaely 0.5, since boh modes are equally srong. The dashed one sandard deviaion lines indicae a decrease in he weigh variabiliy as he number of paricles increases. where B(h 1, h 2 ) is he Bhaacharyya disance beween he normalised N b bin hisograms h 1 and h 2, defined as [ N b ] 1/2 B(h 1, h 2 )= 1 hb,1 h b,2 [0, 1]. b=1 Thus he closer he colour hisograms in he hypohesised bounding box are o he corresponding colour hisograms in he reference bounding box, he higher he likelihood for he hypohesis. The widh of he likelihood is conrolled by he variance parameer σ 2. This likelihood is highly non-linear due o he mapping from he sae o he measuremens. A similar model was employed in he conex of objec racking before in [10]. This model wih N b = 30, σ = 0.15, and (σ 2 x,σ 2 y,σ 2 s x,σ 2 s y ) = (2.5, 1.5, 0.05, 0.05), was used o rack he fooball players in red in he video sequence for which a number of keyframes appear in Figure 3. For boh he sandard and he mixure paricle filers he proposal was aken o be he dynamics, and he iniial paricles were generaed around he red fooball players in he firs frame. For he sandard paricle filer 200 paricles were used, whereas he mixure paricle filer was consrained o have a maximum of 10 componens, wih 20 paricles for each componen alive. Thus he oal number of paricles for he mixure paricle filer is always equal or less han ha for he sandard paricle filer. A ypical racking resul for boh algorihms is depiced in Figure 3. Boh algorihms are iniialised in exacly he same manner around he four players o be racked. Even wih a large number of paricles he sandard paricle filer is unable o mainain he muli-modaliy for more han a few frames. The mixure paricle filer, however, quickly discovers he four main modes, and successfully rack hem hroughou he video sequence. 6. Conclusions ha he corresponding hypohesised bounding box becomes B x =(x, y, s x l x,s y l y ). The variables s x and s y hus ac as scale facors. The componens of he sae are assumed o follow independen Gaussian random walk models wih variances (σ 2 x,σ 2 y,σ 2 s x,σ 2 s y ). The measuremens are aken o be he normalised hisograms of he pixel colour componens wihin he bounding box, i.e. y =(h R B x, h G B x, h B B x ). Noe ha he measuremens depend on he objec sae. The This paper proposed o model he filering disribuion as a mixure model o beer cope wih he muli-modaliy ha may arise due o ambiguiy or he presence of muliple objecs. The general racking recursion for he mixure model was shown o comprise a predicion sep and an updae sep, similar o he sandard Bayesian recursion for a single componen model. I was also shown how a Mone Carlo implemenaion of he general recursion leads o a mixure of paricle filers ha inerac only in he compuaion of he Proceedings of he Ninh IEEE Inernaional Conference on Compuer Vision (ICCV 2003) 2-Volume Se
7 iniialisaion #17 #21 #105 Figure 3. Visual racking resuls. (Top) Sandard paricle filer wih 200 paricles. (Boom) Mixure paricle filer wih a maximum of 10 mixure componens and 20 paricles per componen. All he paricles for he sandard paricle filer quickly migrae o one of he modes. The mixure paricle filer rapidly discovers he four main modes and successfully rack hem hroughou he video sequence. mixure weighs. This mixure paricle filer is able o mainain he muli-modaliy inheren in racking problems where he sandard paricle filer fails, as was illusraed on a synheic and a real world racking problem. References [1] B. D. O. Anderson and J. B. Moore. Opimal Filering. Prenice-Hall, Englewood Cliffs, [2] Y. Bar-Shalom, L. Xiao-Rong, and T. Kirubarajan. Esimaion, Tracking and Navigaion: Theory, Algorihms and Sofware. John Wiley and Sons, [3] S. L. Docksader and A. M. Tekalp. Tracking muliple objecs in he presence of ariculaed and occluded moion. In Proceedings of he IEEE Workshop on Human Moion, pages 88 95, [4] A. Douce, J. F. G. de Freias, and N. J. Gordon, ediors. Sequenial Mone Carlo Mehods in Pracice. Springer-Verlag, New York, [5] A. Douce, S. J. Godsill, and C. Andrieu. On sequenial Mone Carlo sampling mehods for Bayesian filering. Saisics and Compuing, 10(3): , [7] C. Hue, J.-P. Le Cadre, and P. Pérez. Tracking muliple objecs wih paricle filering. IEEE Transacions on Aerospace and Elecronic Sysems, 38(3): , [8] M. Isard and J. MacCormick. BraMBLe: A Bayesian muliple-blob racker. In Proc. In. Conf. Compuer Vision, pages II: 34 41, [9] A.Milsein,J.N.Sánchez, and E. T. Williamson. Robus global localizaion using clusered paricle filering. In Proceedings of AAAI/IAAI, pages , [10] P. Pérez, C. Hue, J. Vermaak, and M. Gangne. Colorbased probabilisic racking. In Proc. Europ. Conf. Compuer Vision, pages I: , [11] C. Rasmussen. Join likelihood mehods for miigaing visual racking disurbances. In Proceedings of he IEEE Workshop on Muli-Objec Tracking, pages 69 76, [12] S. Thrun, D. Fox, W. Burgard, and F. Dellaer. Robus Mone Carlo localizaion for mobile robos. Arificial Inelligence, 128(1-2):00 141, [6] S. Gokurk, J. Bougue, and R. Grzeszczuk. A daadriven model for monocular face racking. In Proc. In. Conf. Compuer Vision, pages , Proceedings of he Ninh IEEE Inernaional Conference on Compuer Vision (ICCV 2003) 2-Volume Se
Real-time Particle Filters
Real-ime Paricle Filers Cody Kwok Dieer Fox Marina Meilă Dep. of Compuer Science & Engineering, Dep. of Saisics Universiy of Washingon Seale, WA 9895 ckwok,fox @cs.washingon.edu, mmp@sa.washingon.edu Absrac
More informationAn Online Learning-based Framework for Tracking
An Online Learning-based Framework for Tracking Kamalika Chaudhuri Compuer Science and Engineering Universiy of California, San Diego La Jolla, CA 9293 Yoav Freund Compuer Science and Engineering Universiy
More informationBayesian Filtering with Online Gaussian Process Latent Variable Models
Bayesian Filering wih Online Gaussian Process Laen Variable Models Yali Wang Laval Universiy yali.wang.1@ulaval.ca Marcus A. Brubaker TTI Chicago mbrubake@cs.orono.edu Brahim Chaib-draa Laval Universiy
More informationA Bayesian framework with auxiliary particle filter for GMTI based ground vehicle tracking aided by domain knowledge
A Bayesian framework wih auxiliary paricle filer for GMTI based ground vehicle racking aided by domain knowledge Miao Yu a, Cunjia Liu a, Wen-hua Chen a and Jonahon Chambers b a Deparmen of Aeronauical
More informationSELF-EVALUATION FOR VIDEO TRACKING SYSTEMS
SELF-EVALUATION FOR VIDEO TRACKING SYSTEMS Hao Wu and Qinfen Zheng Cenre for Auomaion Research Dep. of Elecrical and Compuer Engineering Universiy of Maryland, College Park, MD-20742 {wh2003, qinfen}@cfar.umd.edu
More informationMeasuring macroeconomic volatility Applications to export revenue data, 1970-2005
FONDATION POUR LES ETUDES ET RERS LE DEVELOPPEMENT INTERNATIONAL Measuring macroeconomic volailiy Applicaions o expor revenue daa, 1970-005 by Joël Cariolle Policy brief no. 47 March 01 The FERDI is a
More informationANALYSIS AND COMPARISONS OF SOME SOLUTION CONCEPTS FOR STOCHASTIC PROGRAMMING PROBLEMS
ANALYSIS AND COMPARISONS OF SOME SOLUTION CONCEPTS FOR STOCHASTIC PROGRAMMING PROBLEMS R. Caballero, E. Cerdá, M. M. Muñoz and L. Rey () Deparmen of Applied Economics (Mahemaics), Universiy of Málaga,
More informationParticle Filtering for Geometric Active Contours with Application to Tracking Moving and Deforming Objects
Paricle Filering for Geomeric Acive Conours wih Applicaion o Tracking Moving and Deforming Objecs Yogesh Rahi Namraa Vaswani Allen Tannenbaum Anhony Yezzi Georgia Insiue of Technology School of Elecrical
More informationDuration and Convexity ( ) 20 = Bond B has a maturity of 5 years and also has a required rate of return of 10%. Its price is $613.
Graduae School of Business Adminisraion Universiy of Virginia UVA-F-38 Duraion and Convexiy he price of a bond is a funcion of he promised paymens and he marke required rae of reurn. Since he promised
More informationJournal Of Business & Economics Research September 2005 Volume 3, Number 9
Opion Pricing And Mone Carlo Simulaions George M. Jabbour, (Email: jabbour@gwu.edu), George Washingon Universiy Yi-Kang Liu, (yikang@gwu.edu), George Washingon Universiy ABSTRACT The advanage of Mone Carlo
More informationCombination of UWB and GPS for indoor-outdoor vehicle localization
ombinaion of UW and for indoor-oudoor vehicle localizaion González J., lanco J.L., Galindo., Oriz-de-Galiseo., Fernández-Madrigal J.., Moreno F.., and Marínez J.L. {jgonzalez, jlblanco,cipriano,jafma}@cima.uma.es,
More informationOption Put-Call Parity Relations When the Underlying Security Pays Dividends
Inernaional Journal of Business and conomics, 26, Vol. 5, No. 3, 225-23 Opion Pu-all Pariy Relaions When he Underlying Securiy Pays Dividends Weiyu Guo Deparmen of Finance, Universiy of Nebraska Omaha,
More informationChapter 8: Regression with Lagged Explanatory Variables
Chaper 8: Regression wih Lagged Explanaory Variables Time series daa: Y for =1,..,T End goal: Regression model relaing a dependen variable o explanaory variables. Wih ime series new issues arise: 1. One
More informationTerm Structure of Prices of Asian Options
Term Srucure of Prices of Asian Opions Jirô Akahori, Tsuomu Mikami, Kenji Yasuomi and Teruo Yokoa Dep. of Mahemaical Sciences, Risumeikan Universiy 1-1-1 Nojihigashi, Kusasu, Shiga 525-8577, Japan E-mail:
More informationImproved Techniques for Grid Mapping with Rao-Blackwellized Particle Filters
1 Improved Techniques for Grid Mapping wih Rao-Blackwellized Paricle Filers Giorgio Grisei Cyrill Sachniss Wolfram Burgard Universiy of Freiburg, Dep. of Compuer Science, Georges-Köhler-Allee 79, D-79110
More informationAnalogue and Digital Signal Processing. First Term Third Year CS Engineering By Dr Mukhtiar Ali Unar
Analogue and Digial Signal Processing Firs Term Third Year CS Engineering By Dr Mukhiar Ali Unar Recommended Books Haykin S. and Van Veen B.; Signals and Sysems, John Wiley& Sons Inc. ISBN: 0-7-380-7 Ifeachor
More informationUSE OF EDUCATION TECHNOLOGY IN ENGLISH CLASSES
USE OF EDUCATION TECHNOLOGY IN ENGLISH CLASSES Mehme Nuri GÖMLEKSİZ Absrac Using educaion echnology in classes helps eachers realize a beer and more effecive learning. In his sudy 150 English eachers were
More informationImproving Unreliable Mobile GIS with Swarm-based Particle Filters
Improving Unreliable Mobile GIS wih Swarm-based Paricle Filers Fama Hrizi, Jérôme Härri, Chrisian Bonne EURECOM, Mobile Communicaions Deparmen Campus SophiaTech, 450 Roue des Chappes Bio, France {hrizi,
More informationMarket Liquidity and the Impacts of the Computerized Trading System: Evidence from the Stock Exchange of Thailand
36 Invesmen Managemen and Financial Innovaions, 4/4 Marke Liquidiy and he Impacs of he Compuerized Trading Sysem: Evidence from he Sock Exchange of Thailand Sorasar Sukcharoensin 1, Pariyada Srisopisawa,
More informationINTEREST RATE FUTURES AND THEIR OPTIONS: SOME PRICING APPROACHES
INTEREST RATE FUTURES AND THEIR OPTIONS: SOME PRICING APPROACHES OPENGAMMA QUANTITATIVE RESEARCH Absrac. Exchange-raded ineres rae fuures and heir opions are described. The fuure opions include hose paying
More informationRandom Scanning Algorithm for Tracking Curves in Binary Image Sequences
Vol., No., Page 101 of 110 Copyrigh 008, TSI Press Prined in he USA. All righs reserved Random Scanning Algorihm for Tracking Curves in Binary Image Sequences Kazuhiko Kawamoo *1 and Kaoru Hiroa 1 Kyushu
More informationDYNAMIC MODELS FOR VALUATION OF WRONGFUL DEATH PAYMENTS
DYNAMIC MODELS FOR VALUATION OF WRONGFUL DEATH PAYMENTS Hong Mao, Shanghai Second Polyechnic Universiy Krzyszof M. Osaszewski, Illinois Sae Universiy Youyu Zhang, Fudan Universiy ABSTRACT Liigaion, exper
More informationLoad Prediction Using Hybrid Model for Computational Grid
Load Predicion Using Hybrid Model for Compuaional Grid Yongwei Wu, Yulai Yuan, Guangwen Yang 3, Weimin Zheng 4 Deparmen of Compuer Science and Technology, Tsinghua Universiy, Beijing 00084, China, 3, 4
More informationCHARGE AND DISCHARGE OF A CAPACITOR
REFERENCES RC Circuis: Elecrical Insrumens: Mos Inroducory Physics exs (e.g. A. Halliday and Resnick, Physics ; M. Sernheim and J. Kane, General Physics.) This Laboraory Manual: Commonly Used Insrumens:
More informationARCH 2013.1 Proceedings
Aricle from: ARCH 213.1 Proceedings Augus 1-4, 212 Ghislain Leveille, Emmanuel Hamel A renewal model for medical malpracice Ghislain Léveillé École d acuaria Universié Laval, Québec, Canada 47h ARC Conference
More informationPredicting Stock Market Index Trading Signals Using Neural Networks
Predicing Sock Marke Index Trading Using Neural Neworks C. D. Tilakarane, S. A. Morris, M. A. Mammadov, C. P. Hurs Cenre for Informaics and Applied Opimizaion School of Informaion Technology and Mahemaical
More informationParticle Filtering for Multiple Object Tracking in Dynamic Fluorescence Microscopy Images: Application to Microtubule Growth Analysis
Chaper Three Paricle Filering for Muliple Objec Tracking in Dynamic Fluorescence Microscopy Images: Applicaion o Microubule Growh Analysis I is remarkable ha a science which began wih he consideraion of
More informationPROFIT TEST MODELLING IN LIFE ASSURANCE USING SPREADSHEETS PART ONE
Profi Tes Modelling in Life Assurance Using Spreadshees PROFIT TEST MODELLING IN LIFE ASSURANCE USING SPREADSHEETS PART ONE Erik Alm Peer Millingon 2004 Profi Tes Modelling in Life Assurance Using Spreadshees
More informationMACROECONOMIC FORECASTS AT THE MOF A LOOK INTO THE REAR VIEW MIRROR
MACROECONOMIC FORECASTS AT THE MOF A LOOK INTO THE REAR VIEW MIRROR The firs experimenal publicaion, which summarised pas and expeced fuure developmen of basic economic indicaors, was published by he Minisry
More informationSPEC model selection algorithm for ARCH models: an options pricing evaluation framework
Applied Financial Economics Leers, 2008, 4, 419 423 SEC model selecion algorihm for ARCH models: an opions pricing evaluaion framework Savros Degiannakis a, * and Evdokia Xekalaki a,b a Deparmen of Saisics,
More informationSingle-machine Scheduling with Periodic Maintenance and both Preemptive and. Non-preemptive jobs in Remanufacturing System 1
Absrac number: 05-0407 Single-machine Scheduling wih Periodic Mainenance and boh Preempive and Non-preempive jobs in Remanufacuring Sysem Liu Biyu hen Weida (School of Economics and Managemen Souheas Universiy
More informationThe Belief Roadmap: Efficient Planning in Belief Space by Factoring the Covariance
1 The Belief Roadmap: Efficien Planning in Belief Space by Facoring he Covariance Samuel Prenice and Nicholas Roy Absrac When a mobile agen does no known is posiion perfecly, incorporaing he prediced uncerainy
More informationMaking a Faster Cryptanalytic Time-Memory Trade-Off
Making a Faser Crypanalyic Time-Memory Trade-Off Philippe Oechslin Laboraoire de Securié e de Crypographie (LASEC) Ecole Polyechnique Fédérale de Lausanne Faculé I&C, 1015 Lausanne, Swizerland philippe.oechslin@epfl.ch
More informationTEMPORAL PATTERN IDENTIFICATION OF TIME SERIES DATA USING PATTERN WAVELETS AND GENETIC ALGORITHMS
TEMPORAL PATTERN IDENTIFICATION OF TIME SERIES DATA USING PATTERN WAVELETS AND GENETIC ALGORITHMS RICHARD J. POVINELLI AND XIN FENG Deparmen of Elecrical and Compuer Engineering Marquee Universiy, P.O.
More informationAutomatic measurement and detection of GSM interferences
Auomaic measuremen and deecion of GSM inerferences Poor speech qualiy and dropped calls in GSM neworks may be caused by inerferences as a resul of high raffic load. The radio nework analyzers from Rohde
More informationA Probability Density Function for Google s stocks
A Probabiliy Densiy Funcion for Google s socks V.Dorobanu Physics Deparmen, Poliehnica Universiy of Timisoara, Romania Absrac. I is an approach o inroduce he Fokker Planck equaion as an ineresing naural
More informationRandom Walk in 1-D. 3 possible paths x vs n. -5 For our random walk, we assume the probabilities p,q do not depend on time (n) - stationary
Random Walk in -D Random walks appear in many cones: diffusion is a random walk process undersanding buffering, waiing imes, queuing more generally he heory of sochasic processes gambling choosing he bes
More informationThis document is downloaded from DR-NTU, Nanyang Technological University Library, Singapore.
This documen is downloaded from DR-NTU, Nanyang Technological Universiy Library, Singapore. Tile A Bayesian mulivariae risk-neural mehod for pricing reverse morgages Auhor(s) Kogure, Asuyuki; Li, Jackie;
More informationBayesian model comparison with un-normalised likelihoods
Saisics and Compuing manuscrip No. (will be insered by he edior) Bayesian model comparison wih un-normalised likelihoods Richard G. Everi Adam M. Johansen Ellen Rowing Melina Evdemon-Hogan Each of hese
More informationDOES TRADING VOLUME INFLUENCE GARCH EFFECTS? SOME EVIDENCE FROM THE GREEK MARKET WITH SPECIAL REFERENCE TO BANKING SECTOR
Invesmen Managemen and Financial Innovaions, Volume 4, Issue 3, 7 33 DOES TRADING VOLUME INFLUENCE GARCH EFFECTS? SOME EVIDENCE FROM THE GREEK MARKET WITH SPECIAL REFERENCE TO BANKING SECTOR Ahanasios
More informationA PROPOSAL TO OBTAIN A LONG QUARTERLY CHILEAN GDP SERIES *
CUADERNOS DE ECONOMÍA, VOL. 43 (NOVIEMBRE), PP. 285-299, 2006 A PROPOSAL TO OBTAIN A LONG QUARTERLY CHILEAN GDP SERIES * JUAN DE DIOS TENA Universidad de Concepción y Universidad Carlos III, España MIGUEL
More information11/6/2013. Chapter 14: Dynamic AD-AS. Introduction. Introduction. Keeping track of time. The model s elements
Inroducion Chaper 14: Dynamic D-S dynamic model of aggregae and aggregae supply gives us more insigh ino how he economy works in he shor run. I is a simplified version of a DSGE model, used in cuing-edge
More informationRisk Modelling of Collateralised Lending
Risk Modelling of Collaeralised Lending Dae: 4-11-2008 Number: 8/18 Inroducion This noe explains how i is possible o handle collaeralised lending wihin Risk Conroller. The approach draws on he faciliies
More informationWhy Did the Demand for Cash Decrease Recently in Korea?
Why Did he Demand for Cash Decrease Recenly in Korea? Byoung Hark Yoo Bank of Korea 26. 5 Absrac We explores why cash demand have decreased recenly in Korea. The raio of cash o consumpion fell o 4.7% in
More informationMorningstar Investor Return
Morningsar Invesor Reurn Morningsar Mehodology Paper Augus 31, 2010 2010 Morningsar, Inc. All righs reserved. The informaion in his documen is he propery of Morningsar, Inc. Reproducion or ranscripion
More informationMeasuring the Effects of Monetary Policy: A Factor-Augmented Vector Autoregressive (FAVAR) Approach * Ben S. Bernanke, Federal Reserve Board
Measuring he Effecs of Moneary Policy: A acor-augmened Vecor Auoregressive (AVAR) Approach * Ben S. Bernanke, ederal Reserve Board Jean Boivin, Columbia Universiy and NBER Pior Eliasz, Princeon Universiy
More informationTime Series Analysis Using SAS R Part I The Augmented Dickey-Fuller (ADF) Test
ABSTRACT Time Series Analysis Using SAS R Par I The Augmened Dickey-Fuller (ADF) Tes By Ismail E. Mohamed The purpose of his series of aricles is o discuss SAS programming echniques specifically designed
More informationSupplementary Appendix for Depression Babies: Do Macroeconomic Experiences Affect Risk-Taking?
Supplemenary Appendix for Depression Babies: Do Macroeconomic Experiences Affec Risk-Taking? Ulrike Malmendier UC Berkeley and NBER Sefan Nagel Sanford Universiy and NBER Sepember 2009 A. Deails on SCF
More informationOptimal Investment and Consumption Decision of Family with Life Insurance
Opimal Invesmen and Consumpion Decision of Family wih Life Insurance Minsuk Kwak 1 2 Yong Hyun Shin 3 U Jin Choi 4 6h World Congress of he Bachelier Finance Sociey Torono, Canada June 25, 2010 1 Speaker
More informationHow To Calculate Price Elasiciy Per Capia Per Capi
Price elasiciy of demand for crude oil: esimaes for 23 counries John C.B. Cooper Absrac This paper uses a muliple regression model derived from an adapaion of Nerlove s parial adjusmen model o esimae boh
More informationChapter 7. Response of First-Order RL and RC Circuits
Chaper 7. esponse of Firs-Order L and C Circuis 7.1. The Naural esponse of an L Circui 7.2. The Naural esponse of an C Circui 7.3. The ep esponse of L and C Circuis 7.4. A General oluion for ep and Naural
More informationAnalysis of Planck and the Equilibrium ofantis in Tropical Physics
Emergence of Fokker-Planck Dynamics wihin a Closed Finie Spin Sysem H. Niemeyer(*), D. Schmidke(*), J. Gemmer(*), K. Michielsen(**), H. de Raed(**) (*)Universiy of Osnabrück, (**) Supercompuing Cener Juelich
More informationPrincipal components of stock market dynamics. Methodology and applications in brief (to be updated ) Andrei Bouzaev, bouzaev@ya.
Principal componens of sock marke dynamics Mehodology and applicaions in brief o be updaed Andrei Bouzaev, bouzaev@ya.ru Why principal componens are needed Objecives undersand he evidence of more han one
More informationPerformance Center Overview. Performance Center Overview 1
Performance Cener Overview Performance Cener Overview 1 ODJFS Performance Cener ce Cener New Performance Cener Model Performance Cener Projec Meeings Performance Cener Execuive Meeings Performance Cener
More informationMultiprocessor Systems-on-Chips
Par of: Muliprocessor Sysems-on-Chips Edied by: Ahmed Amine Jerraya and Wayne Wolf Morgan Kaufmann Publishers, 2005 2 Modeling Shared Resources Conex swiching implies overhead. On a processing elemen,
More informationHow To Predict A Person'S Behavior
Informaion Theoreic Approaches for Predicive Models: Resuls and Analysis Monica Dinculescu Supervised by Doina Precup Absrac Learning he inernal represenaion of parially observable environmens has proven
More informationDoes Option Trading Have a Pervasive Impact on Underlying Stock Prices? *
Does Opion Trading Have a Pervasive Impac on Underlying Sock Prices? * Neil D. Pearson Universiy of Illinois a Urbana-Champaign Allen M. Poeshman Universiy of Illinois a Urbana-Champaign Joshua Whie Universiy
More informationA Natural Feature-Based 3D Object Tracking Method for Wearable Augmented Reality
A Naural Feaure-Based 3D Objec Tracking Mehod for Wearable Augmened Realiy Takashi Okuma Columbia Universiy / AIST Email: okuma@cs.columbia.edu Takeshi Kuraa Universiy of Washingon / AIST Email: kuraa@ieee.org
More informationII.1. Debt reduction and fiscal multipliers. dbt da dpbal da dg. bal
Quarerly Repor on he Euro Area 3/202 II.. Deb reducion and fiscal mulipliers The deerioraion of public finances in he firs years of he crisis has led mos Member Saes o adop sizeable consolidaion packages.
More informationLarge Scale Online Learning.
Large Scale Online Learning. Léon Boou NEC Labs America Princeon NJ 08540 leon@boou.org Yann Le Cun NEC Labs America Princeon NJ 08540 yann@lecun.com Absrac We consider siuaions where raining daa is abundan
More informationStochastic Optimal Control Problem for Life Insurance
Sochasic Opimal Conrol Problem for Life Insurance s. Basukh 1, D. Nyamsuren 2 1 Deparmen of Economics and Economerics, Insiue of Finance and Economics, Ulaanbaaar, Mongolia 2 School of Mahemaics, Mongolian
More informationTime Varying Coefficient Models; A Proposal for selecting the Coefficient Driver Sets
Time Varying Coefficien Models; A Proposal for selecing he Coefficien Driver Ses Sephen G. Hall, Universiy of Leiceser P. A. V. B. Swamy George S. Tavlas, Bank of Greece Working Paper No. 14/18 December
More informationDynamic programming models and algorithms for the mutual fund cash balance problem
Submied o Managemen Science manuscrip Dynamic programming models and algorihms for he muual fund cash balance problem Juliana Nascimeno Deparmen of Operaions Research and Financial Engineering, Princeon
More informationThe naive method discussed in Lecture 1 uses the most recent observations to forecast future values. That is, Y ˆ t + 1
Business Condiions & Forecasing Exponenial Smoohing LECTURE 2 MOVING AVERAGES AND EXPONENTIAL SMOOTHING OVERVIEW This lecure inroduces ime-series smoohing forecasing mehods. Various models are discussed,
More informationHotel Room Demand Forecasting via Observed Reservation Information
Proceedings of he Asia Pacific Indusrial Engineering & Managemen Sysems Conference 0 V. Kachivichyanuul, H.T. Luong, and R. Piaaso Eds. Hoel Room Demand Forecasing via Observed Reservaion Informaion aragain
More informationA New Adaptive Ensemble Boosting Classifier for Concept Drifting Stream Data
A New Adapive Ensemble Boosing Classifier for Concep Drifing Sream Daa Kapil K. Wankhade and Snehlaa S. Dongre, Members, IACSIT Absrac Wih he emergence of large-volume and high speed sreaming daa, mining
More informationStatistical Analysis with Little s Law. Supplementary Material: More on the Call Center Data. by Song-Hee Kim and Ward Whitt
Saisical Analysis wih Lile s Law Supplemenary Maerial: More on he Call Cener Daa by Song-Hee Kim and Ward Whi Deparmen of Indusrial Engineering and Operaions Research Columbia Universiy, New York, NY 17-99
More informationPremium Income of Indian Life Insurance Industry
Premium Income of Indian Life Insurance Indusry A Toal Facor Produciviy Approach Ram Praap Sinha* Subsequen o he passage of he Insurance Regulaory and Developmen Auhoriy (IRDA) Ac, 1999, he life insurance
More informationThe Application of Multi Shifts and Break Windows in Employees Scheduling
The Applicaion of Muli Shifs and Brea Windows in Employees Scheduling Evy Herowai Indusrial Engineering Deparmen, Universiy of Surabaya, Indonesia Absrac. One mehod for increasing company s performance
More informationE0 370 Statistical Learning Theory Lecture 20 (Nov 17, 2011)
E0 370 Saisical Learning Theory Lecure 0 (ov 7, 0 Online Learning from Expers: Weighed Majoriy and Hedge Lecurer: Shivani Agarwal Scribe: Saradha R Inroducion In his lecure, we will look a he problem of
More informationBehavior Analysis of a Biscuit Making Plant using Markov Regenerative Modeling
Behavior Analysis of a Biscui Making lan using Markov Regeneraive Modeling arvinder Singh & Aul oyal Deparmen of Mechanical Engineering, Lala Lajpa Rai Insiue of Engineering & Technology, Moga -, India
More informationA Distributed Multiple-Target Identity Management Algorithm in Sensor Networks
A Disribued Muliple-Targe Ideniy Managemen Algorihm in Sensor Neworks Inseok Hwang, Kaushik Roy, Hamsa Balakrishnan, and Claire Tomlin Dep. of Aeronauics and Asronauics, Sanford Universiy, CA 94305 Elecrical
More informationDETERMINISTIC INVENTORY MODEL FOR ITEMS WITH TIME VARYING DEMAND, WEIBULL DISTRIBUTION DETERIORATION AND SHORTAGES KUN-SHAN WU
Yugoslav Journal of Operaions Research 2 (22), Number, 6-7 DEERMINISIC INVENORY MODEL FOR IEMS WIH IME VARYING DEMAND, WEIBULL DISRIBUION DEERIORAION AND SHORAGES KUN-SHAN WU Deparmen of Bussines Adminisraion
More informationON THE PRICING OF EQUITY-LINKED LIFE INSURANCE CONTRACTS IN GAUSSIAN FINANCIAL ENVIRONMENT
Teor Imov r.amaem.sais. Theor. Probabiliy and Mah. Sais. Vip. 7, 24 No. 7, 25, Pages 15 111 S 94-9(5)634-4 Aricle elecronically published on Augus 12, 25 ON THE PRICING OF EQUITY-LINKED LIFE INSURANCE
More informationThe Grantor Retained Annuity Trust (GRAT)
WEALTH ADVISORY Esae Planning Sraegies for closely-held, family businesses The Granor Reained Annuiy Trus (GRAT) An efficien wealh ransfer sraegy, paricularly in a low ineres rae environmen Family business
More informationInformation Theoretic Evaluation of Change Prediction Models for Large-Scale Software
Informaion Theoreic Evaluaion of Change Predicion Models for Large-Scale Sofware Mina Askari School of Compuer Science Universiy of Waerloo Waerloo, Canada maskari@uwaerloo.ca Ric Hol School of Compuer
More informationStochastic Recruitment: A Limited-Feedback Control Policy for Large Ensemble Systems
Sochasic Recruimen: A Limied-Feedback Conrol Policy for Large Ensemble Sysems Lael Odhner and Harry Asada Absrac This paper is abou sochasic recruimen, a conrol archiecure for cenrally conrolling he ensemble
More informationadaptive control; stochastic systems; certainty equivalence principle; long-term
COMMUICATIOS I IFORMATIO AD SYSTEMS c 2006 Inernaional Press Vol. 6, o. 4, pp. 299-320, 2006 003 ADAPTIVE COTROL OF LIEAR TIME IVARIAT SYSTEMS: THE BET O THE BEST PRICIPLE S. BITTATI AD M. C. CAMPI Absrac.
More informationDistributing Human Resources among Software Development Projects 1
Disribuing Human Resources among Sofware Developmen Proecs Macario Polo, María Dolores Maeos, Mario Piaini and rancisco Ruiz Summary This paper presens a mehod for esimaing he disribuion of human resources
More informationMathematics in Pharmacokinetics What and Why (A second attempt to make it clearer)
Mahemaics in Pharmacokineics Wha and Why (A second aemp o make i clearer) We have used equaions for concenraion () as a funcion of ime (). We will coninue o use hese equaions since he plasma concenraions
More informationMTH6121 Introduction to Mathematical Finance Lesson 5
26 MTH6121 Inroducion o Mahemaical Finance Lesson 5 Conens 2.3 Brownian moion wih drif........................... 27 2.4 Geomeric Brownian moion........................... 28 2.5 Convergence of random
More informationDistributed Online Localization in Sensor Networks Using a Moving Target
Disribued Online Localizaion in Sensor Neworks Using a Moving Targe Aram Galsyan 1, Bhaskar Krishnamachari 2, Krisina Lerman 1, and Sundeep Paem 2 1 Informaion Sciences Insiue 2 Deparmen of Elecrical Engineering-Sysems
More informationThe Kinetics of the Stock Markets
Asia Pacific Managemen Review (00) 7(1), 1-4 The Kineics of he Sock Markes Hsinan Hsu * and Bin-Juin Lin ** (received July 001; revision received Ocober 001;acceped November 001) This paper applies he
More informationCOMPARISON OF AIR TRAVEL DEMAND FORECASTING METHODS
COMPARISON OF AIR RAVE DEMAND FORECASING MEHODS Ružica Škurla Babić, M.Sc. Ivan Grgurević, B.Eng. Universiy of Zagreb Faculy of ranspor and raffic Sciences Vukelićeva 4, HR- Zagreb, Croaia skurla@fpz.hr,
More informationReal Time Bid Optimization with Smooth Budget Delivery in Online Advertising
Real Time Bid Opimizaion wih Smooh Budge Delivery in Online Adverising Kuang-Chih Lee Ali Jalali Ali Dasdan Turn Inc. 835 Main Sree, Redwood Ciy, CA 94063 {klee,ajalali,adasdan}@urn.com ABSTRACT Today,
More informationIndividual Health Insurance April 30, 2008 Pages 167-170
Individual Healh Insurance April 30, 2008 Pages 167-170 We have received feedback ha his secion of he e is confusing because some of he defined noaion is inconsisen wih comparable life insurance reserve
More informationInventory Planning with Forecast Updates: Approximate Solutions and Cost Error Bounds
OPERATIONS RESEARCH Vol. 54, No. 6, November December 2006, pp. 1079 1097 issn 0030-364X eissn 1526-5463 06 5406 1079 informs doi 10.1287/opre.1060.0338 2006 INFORMS Invenory Planning wih Forecas Updaes:
More informationChapter 1.6 Financial Management
Chaper 1.6 Financial Managemen Par I: Objecive ype quesions and answers 1. Simple pay back period is equal o: a) Raio of Firs cos/ne yearly savings b) Raio of Annual gross cash flow/capial cos n c) = (1
More informationAP Calculus BC 2010 Scoring Guidelines
AP Calculus BC Scoring Guidelines The College Board The College Board is a no-for-profi membership associaion whose mission is o connec sudens o college success and opporuniy. Founded in, he College Board
More informationStrategic Optimization of a Transportation Distribution Network
Sraegic Opimizaion of a Transporaion Disribuion Nework K. John Sophabmixay, Sco J. Mason, Manuel D. Rossei Deparmen of Indusrial Engineering Universiy of Arkansas 4207 Bell Engineering Cener Fayeeville,
More informationBiology at Home - Pariion Funcion Guillaume
On Tracking The Pariion Funcion Guillaume Desjardins, Aaron Courville, Yoshua Bengio {desjagui,courvila,bengioy}@iro.umonreal.ca Déparemen d informaique e de recherche opéraionnelle Universié de Monréal
More informationPricing Fixed-Income Derivaives wih he Forward-Risk Adjused Measure Jesper Lund Deparmen of Finance he Aarhus School of Business DK-8 Aarhus V, Denmark E-mail: jel@hha.dk Homepage: www.hha.dk/~jel/ Firs
More informationAPPLICATION OF THE KALMAN FILTER FOR ESTIMATING CONTINUOUS TIME TERM STRUCTURE MODELS: THE CASE OF UK AND GERMANY. January, 2005
APPLICATION OF THE KALMAN FILTER FOR ESTIMATING CONTINUOUS TIME TERM STRUCTURE MODELS: THE CASE OF UK AND GERMANY Somnah Chaeree* Deparmen of Economics Universiy of Glasgow January, 2005 Absrac The purpose
More informationCan Individual Investors Use Technical Trading Rules to Beat the Asian Markets?
Can Individual Invesors Use Technical Trading Rules o Bea he Asian Markes? INTRODUCTION In radiional ess of he weak-form of he Efficien Markes Hypohesis, price reurn differences are found o be insufficien
More informationChapter 2 Kinematics in One Dimension
Chaper Kinemaics in One Dimension Chaper DESCRIBING MOTION:KINEMATICS IN ONE DIMENSION PREVIEW Kinemaics is he sudy of how hings moe how far (disance and displacemen), how fas (speed and elociy), and how
More informationForecasting Product Sales with Dynamic Linear Mixture Models. Phillip M. Yelland and Eunice Lee
Forecasing Produc Sales wih Dynamic Linear Mixure Models Phillip M. Yelland and Eunice Lee Forecasing Produc Sales wih Dynamic Linear Mixure Models Phillip M. Yelland Eunice Lee SMLI TR-2003-122 March
More informationTarget Tracking Performance Evaluation A General Software Environment for Filtering
Targe Tracking Performance Evaluaion A General Sofware Environmen for Filering Gusaf Hendeby and Rickard Karlsson Auomaic Conrol, Deparmen of Elecrical Engineering Linköping Universiy SE-581 83 Linköping,
More informationAppendix D Flexibility Factor/Margin of Choice Desktop Research
Appendix D Flexibiliy Facor/Margin of Choice Deskop Research Cheshire Eas Council Cheshire Eas Employmen Land Review Conens D1 Flexibiliy Facor/Margin of Choice Deskop Research 2 Final Ocober 2012 \\GLOBAL.ARUP.COM\EUROPE\MANCHESTER\JOBS\200000\223489-00\4
More informationModule 4. Single-phase AC circuits. Version 2 EE IIT, Kharagpur
Module 4 Single-phase A circuis ersion EE T, Kharagpur esson 5 Soluion of urren in A Series and Parallel ircuis ersion EE T, Kharagpur n he las lesson, wo poins were described:. How o solve for he impedance,
More information