Parameter-Free Convex Learning through Coin Betting

Size: px
Start display at page:

Download "Parameter-Free Convex Learning through Coin Betting"

Transcription

1 JMLR: Workshop and Conference Proceedings 1:1 7, 2016 ICML 2016 AuoML Workshop Parameer-Free Convex Learning hrough Coin Being Francesco Orabona Dávid Pál Yahoo Research, New York Absrac We presen a new parameer-free algorihm for online linear opimizaion over any Hilber space. I is heoreically opimal, wih regre guaranees as good as wih he bes possible learning rae. The algorihm is simple and easy o implemen. The analysis is given via he adversarial coin-being game, Kelly being and he Krichevsky-Trofimov esimaor. Applicaions o obain parameer-free convex opimizaion and machine learning algorihms are shown. 1. Inroducion We consider he Online Linear Opimizaion (OLO) (Cesa-Bianchi and Lugosi, 2006; Shalev-Shwarz, 2011) over a Hilber space H. In each round, an algorihm chooses a poin w H and hen receives a loss vecor l H. The algorihm s goal is o keep is regre small, defined as he difference beween is cumulaive reward and he cumulaive reward of a fixed sraegy u H, ha is Regre T (u) = T l, w =1 T l, u. where, is he inner produc in H. OLO is a basic building block of many machine learning problems. For example, Online Convex Opimizaion (OCO) is a problem analogous o OLO where he linear funcion u l, u is generalized o an arbirary convex funcion f (u). OCO is solved hrough a reducion o OLO by feeding he algorihm l = f (w ) (Shalev-Shwarz, 2011). Bach and sochasic convex opimizaion can also be solved hrough a reducion o OLO by aking he average of w 1, w 2,..., w T (Shalev- Shwarz, 2011). To achieve opimal regre, mos of he exising online algorihms (e.g. Online Gradien Descen, Hedge) require he user o se he learning rae o an unknown/oracle value. Recenly, new parameer-free algorihms have been proposed for OLO/OCO (Chaudhuri e al., 2009; Chernov and Vovk, 2010; Sreeer and McMahan, 2012; Orabona, 2013; McMahan and Abernehy, 2013; McMahan and Orabona, 2014; Luo and Schapire, 2014; Orabona, 2014; Luo and Schapire, 2015; Koolen and van Erven, 2015). These algorihms adap o he characerisics of he opimal predicor, wihou he need o une parameers. However, heir design and underlying inuiion is sill a challenge. Our conribuions are as follows. We connec algorihms for OLO wih coin being. Namely, we show ha an algorihm for OLO can be viewed as an algorihm for being on oucomes of adversarial coin flips. The wealh he algorihm can generae for he being problem is conneced o he regre in OLO seing. This insigh allows us o design novel parameer-free algorihms, =1 c 2016 F. Orabona & D. Pál.

2 ORABONA PÁL which are exremely simple and naural. We also show some applicaions of our resuls o convex opimizaion and machine learning, as well as some empirical resuls. 2. How o Tune he Learning Raes? Denoe by =, he induced norm in H and assume ha l 1. Consider OLO over a Hilber Space H. Online Gradien Descen (OGD) wih learning rae η saisfies (Shalev-Shwarz, 2011) u H Regre T (u) u 2 2η + ηt 2. (1) I is obvious ha he opimal uning of he learning rae depends on he unknown norm of u. The simple choice η = 1/ T leads o an algorihm ha saisfies ( u H Regre T (u) u 2) T. (2) However, in his bound he dependency on u is subopimal: The quadraic dependency can be replaced by an (almos) linear dependency. Saring from (1), if we choose he learning rae η = D/ T, we ge a family of algorihms parameerized by D [0, ) ha saisfy u H : u D = Regre T (u) D T. (3) Insead of a family of algorihms parameerized by D [0, ) saisfying he bound (3), one would like o have a single algorihm (wihou any uning parameers) saisfying u H Regre T (u) u T. (4) Noice ha (4) is sronger han (3) in he following sense: A single algorihm saisfying (4) implies (3) for all values of D [0, ). However, a family of algorihms {A D : D [0, )} parameerized by D where A D saisfies (3), does no yield a single algorihm ha saisfies (4). Finally, noe ha (4) has beer dependency on u han (2). Beer guaranees are indeed possible. In fac, here have been a lo of work on algorihms (Sreeer and McMahan, 2012; Orabona, 2013; McMahan and Abernehy, 2013; McMahan and Orabona, 2014; Orabona, 2014) ha saisfy a slighly weaker version of (4). Namely, heir regre saisfies u H Regre T (u) ( O(1) + polylog(1 + u ) u ) T. (5) I can be shown ha for OLO over Hilber space he exra poly-logarihmic facor is necessary (McMahan and Abernehy, 2013; Orabona, 2013). Algorihms saisfying (5) are called parameer-free, since hey do no need o know D, ye hey have an opimal dependency on u. 3. Parameer-Free Algorihm From Coin Being Here, we presen our new parameer-free algorihm for OLO over a Hilber space H, saed as Algorihm 1. We would like o sress he exreme simpliciy of he algorihm. The heorem below upper bounds is regre in he form of (5), he proof can be found in Orabona and Pál (2016). 2

3 PARAMETER-FREE CONVEX LEARNING THROUGH COIN BETTING Algorihm 1 Algorihm for OLO over Hilber space H based on Krichevsky-Trofimov esimaor 1: for = 1, 2,... do ( 2: Predic wih w 1 1 ) 1 l 1 i, w i l i 3: Receive loss vecor l H such ha l 1 4: end for Theorem 1 (Regre Bound for Algorihm 1) Le {l } =1 be any sequence of loss vecors in a Hilber space H such ha l 1. Algorihm 1 saisfies ( T 0 u H Regre T (u) u T ln 1 + 4T 2 u 2) + 1. We now explain how Algorihm 1 is derived from he Krichevsky-Trofimov soluion o he adversarial coin-being problem. Adversarial Coin Being. Consider a gambler making repeaed bes on he oucomes of adversarial coin flips. The gambler sars wih an iniial endowmen of 1 dollar. In each round, he bes on he oucome of a coin flip c { 1, 1}, where +1 denoes heads and 1 denoes ails. The oucome c is chosen by an adversary. The gambler can be any amoun on eiher heads or ails. However, he canno borrow any addiional money. If he loses, he loses he beed amoun; if he wins, he ges he beed amoun back and, in addiion o ha, he ges he same amoun as a reward. We encode he gambler s be in round by a single number β [ 1, 1]. The sign of β encodes wheher he is being on heads or ails. The absolue value encodes he beed amoun as he fracion of his curren wealh. Le Wealh be gambler s wealh a he end of round. I saisfies Wealh 0 = 1 and Wealh = (1 + c β ) Wealh 1 for 1. (6) Noe ha since β [ 1, 1], gambler s wealh says always non-negaive. Kelly Being and Krichevsky-Trofimov Esimaor. For sequenial being on i.i.d. coin flips, he opimal sraegy has been proposed by Kelly (1956). The sraegy assumes ha he coin flips {c } =1, c {+1, 1}, are generaed i.i.d. wih known probabiliy of heads. If p [0, 1] is he probabiliy of heads, he Kelly be is β = 2p 1. He showed ha, in he long run, his sraegy will provide more wealh han being any oher fixed fracion (Kelly, 1956). For adversarial coins, Kelly being does no make sense. Krichevsky and Trofimov (1981) proposed o replace p wih an esimae: Afer seeing coin flips c 1, c 2,..., c 1, use he empirical esimae k = 1/2+ 1 1[c i=+1]. Their esimae is commonly called KT esimaor 1 and i resuls in 1 he being sraegy β = 2k 1 = c i. Krichevsky and Trofimov showed ha his sraegy guaranees almos he same wealh ha one would obain knowing in advance he fracion of heads. Namely, if we denoe by Wealh (β) he wealh of he sraegy ha bes he fracion β in every round, hen he wealh of he Krichevsky-Trofimov being sraegy saisfies β [ 1, 1] Wealh Wealh (β) 2. (7) Moreover, his guaranee is opimal up o consan muliplicaive facors (Cesa-Bianchi and Lugosi, 2006). 1. Compared o he sandard maximum likelihood esimae 1 1[c i=+1] 1, KT esimaor shrinks slighly owards

4 ORABONA PÁL Algorihm 2 SGD algorihm based on KT esimaor Require: Convex funcions f 1, f 2,..., f N and desired number of ieraions T 1: Iniialize Wealh 0 1 and θ 0 0 2: for = 1, 2,..., T do 3: Se w Wealh 1 θ 1 4: Selec an index j a random from {1, 2,..., N} and compue l = f j (w 1 ) 5: Updae θ θ 1 l and Wealh Wealh 1 l, w 6: end for 7: Oupu w T = 1 T T =1 w From being o OLO. In Algorihm 1, he coin oucome is he vecor c H where c = l and algorihm s wealh is Wealh = 1+ c i, w i = 1 l i, w i. The algorihm explicily 1 keeps rack of is wealh and i bes vecorial fracion β = c 1 i = l i of is curren wealh. The regre bound (Theorem 1) is a consequence of Krichevsky-Trofimov lower bound (7) on he wealh and he dualiy beween regre and wealh. For more deails, see Orabona and Pál (2016). 4. From Online Learning o Convex Opimizaion and Machine Learning The resul in Secion 3 immediaely implies new algorihms and resuls in convex opimizaion and machine learning. We will sae some of hem here, see Orabona (2014) for more resuls. Convex Opimizaion. Consider an empirical risk minimizaion problem of he form F (w) = 1 N N f i (w), (8) where f i : R d R is convex. 2 I is immediae o ransform Algorihm 1 ino a Sochasic Gradien Descen (SGD) algorihm for his problem, obaining Algorihm 2. In Algorihm 2, f j (w) denoes a subgradien of f j a a poin w. We assume ha he norm of he subgradien of f j is bounded by 1. Beside he simpliciy of he Algorihm 2, i has he imporan propery is ha i does no have a learning rae o be uned, ye i achieves he opimal convergence rae. In fac, denoing by ŵ = arg min w F (w) he opimal soluion of (8), he following heorem saes he rae of convergence of Algorihm 2. Theorem 2 The average w T produced by Algorihm 2 is an approximae minimizer of he objecive funcion (8): E [F (w T )] F (ŵ) ŵ T log(1 + 4T 2 ŵ 2 ) + 1 T. Noe ha in he above heorem, T can be larger (muliple epochs) or smaller han N. Machine Learning. In machine learning, he minimizaion of a funcion (8) is jus a proxy o minimize he rue risk over an unknown disribuion. For example, f i (w) can be of he form f i (w) = f(w, X i, Y i ) where {(X i, Y i )} N is a sequences of labeled sampled generaed i.i.d. from some unknown disribuion and f(w, X i, Y i ) is he logisic loss of a weigh vecor w on a sample 2. The algorihm can also be implemened and analyzed wih kernels (Orabona, 2014). 4

5 PARAMETER-FREE CONVEX LEARNING THROUGH COIN BETTING Algorihm 3 Averaging algorihm based on KT esimaor Require: Sample (X 1, Y 1 ), (X 2, Y 2 ),..., (X N, Y N ) 1: Iniialize Wealh 0 1 and θ 0 0 2: for i = 1, 2,..., N do θ 3: Se w i Wealh i 1 i 1 i 4: Compue l i = f(w,x i,y i ) w w=wi 5: Updae θ i θ i 1 l i and Wealh i Wealh i 1 l i, w i 6: end for 7: Oupu w N = 1 N N w i (X i, Y i ). A common approach o have a small risk on he es se is o minimize a regularized objecive funcion over he raining se: F Reg (w) = w N N f(w, X i, Y i ). (9) This problem is srongly convex, so here are very efficien mehods o minimize i, hence we can assume o be able o ge he minimizer of F Reg wih arbirary high precision and algorihms ha do no require o une learning raes. Ye, his is no enough. In fac, we are rarely ineresed in he value of he objecive funcion F Reg or is minimizer, raher we are ineresed in he rue risk of a soluion w, ha is E[f(w, X, Y )], where (X, Y ) is an independen es sample from he same disribuion from which he raining se {(X i, Y i )} N came from. Hence, in order o ge a good performance we have o selec a good regularizaion parameer. In paricular, from Sridharan e al. (2009) we ge E[f(ŵ, X, Y )] E[f(w, X, Y )] O( w N ), where w = arg min w E[f(w, X, Y )] and ŵ = arg min w F Reg (w). From he above bound, i is clear ha he opimal value of depends on he w ha is unknown. Ye anoher possibiliy is o selec he opimal learning rae and/or he number of epochs of SGD o direcly minimize E[f(w, X, Y )]. However, all hese mehods are equivalen (J. Lin, 2016) and hey sill require o une a leas one parameer. We would like o sress ha his is no jus a heoreical problem: Any praciioner knows how painful i is o find he righ regularizaion for he problem a hand. Assuming we would know w, we could se = O(1/( w N)) o achieve he wors-case opimal bound ( ) E[f(ŵ, X, Y )] E[f(w, X, Y )] O w N. (10) However, we can ge he same guaranee wihou knowing w or he opimal, by doing a single pass over he daa se. More precisely, we derive Algorihm 3 from Algorihm 1 by applying he sandard online-o-bach reducion (Shalev-Shwarz, 2011). The algorihm makes only a single pass over he daase and i does no have any uning parameers. Ye, i has almos he same guaranee (10) wihou knowing w or he opimal regularizaion parameer or he learning rae, or any oher uning parameer. 5

6 ORABONA PÁL 0.7 yearpredicionmsd daase, absolue loss 16 cpusmall daase, absolue loss 1.2 x 105 cadaa daase, absolue loss SGD KT-based 1.15 Tes loss Tes loss Tes loss SGD KT-based SGD KT-based Learning rae SGD (a) Learning rae SGD (b) Learning rae SGD (c) Figure 1: Tes loss versus learning rae parameer of SGD (in log scale), compared wih he parameer-free Algorihm 3. Theorem 3 Assume ha (X, Y ), (X 1, Y 1 ), (X 2, Y 2 ),..., (X N, Y N ) are i.i.d. The oupu w N of Algorihm 3 saisfies E[f(w N, X, Y )] E[f(w, X, Y )] w N log(1 + 4N 2 w 2 ) + 1 N. Comparing his guaranee o he one in (10), we see ha, jus paying a sub-logarihmic price, we obain he opimal convergence rae and we remove all he parameers. 5. Empirical Evaluaion We have also run a small empirical evaluaion o show ha he heoreical difference beween classic learning algorihms and parameer-free ones is real and no jus heoreical. In Figure 1, we have used hree regression daases, 3 and solved he OCO problem hrough OLO. In all he hree cases, we have used he absolue loss and normalized he inpu vecors o have L2 norm equal o 1. The daase were spli in wo pars: 75% raining se and he remaining as es se. The raining is done hrough one pass over he raining se and he final classifier is evaluaed on he es se. We used 5 differen splis of raining/es and we repor average and sandard deviaions. We have run SGD wih differen learning raes and shown he performance of is las soluion on he es se. For Algorihm 3, we do no have any parameer o une so we jus plo is es se performance as a line. From he empirical resuls, i is clear ha he opimal learning rae is compleely daa-dependen. I is also ineresing o noe how he performance of SGD becomes very unsable wih large learning raes. Ye our parameer-free algorihm has a performance very close o he unknown opimal uning of he learning rae of SGD. Acknowledgmens The auhors hank Jacob Abernehy, Nicolò Cesa-Bianchi, Sayen Kale, Chansoo Lee, and Giuseppe Moleni for useful discussions on his work. 3. Daases available a hps:// cjlin/libsvmools/daases/. 6

7 PARAMETER-FREE CONVEX LEARNING THROUGH COIN BETTING References N. Cesa-Bianchi and G. Lugosi. Predicion, learning, and games. Cambridge Universiy Press, K. Chaudhuri, Y. Freund, and D. Hsu. A parameer-free hedging algorihm. In Y. Bengio, D. Schuurmans, J. D. Laffery, C. K. I. Williams, and A. Culoa, ediors, Advances in Neural Informaion Processing Sysems 22 (NIPS 2009), December 7-12, Vancouver, BC, Canada, pages , A. Chernov and V. Vovk. Predicion wih advice of unknown number of expers. In Peer Grünwald and Peer Spires, ediors, Proc. of he 26h Conf. on Uncerainy in Arificial Inelligence (UAI 2010), July 8-11, Caalina Island, California, USA. AUAI Press, L. Rosasco J. Lin, R. Camoriano. Generalizaion properies and implici regularizaion for muliple passes SGM. In Proc. of Inernaional Conference on Machine Learning (ICML), J. L. Kelly. A new inerpreaion of informaion rae. Informaion Theory, IRE Transacions on, 2(3): , Sepember W. M. Koolen and T. van Erven. Second-order quanile mehods for expers and combinaorial games. In Peer Grünwald, Elad Hazan, and Sayen Kale, ediors, Proc. of he 28h Conf. on Learning Theory (COLT 2015), July 3-6, Paris, France, pages , R. E. Krichevsky and V. K. Trofimov. The performance of universal encoding. IEEE Transacions on Informaion Theory, 27(2): , H. Luo and R. E. Schapire. A drifing-games analysis for online learning and applicaions o boosing. In Z. Ghahramani, M. Welling, C. Cores, N.D. Lawrence, and K.Q. Weinberger, ediors, Advances in Neural Informaion Processing Sysems 27, pages Curran Associaes, Inc., H. Luo and R. E. Schapire. Achieving all wih no parameers: AdaNormalHedge. In Peer Grünwald, Elad Hazan, and Sayen Kale, ediors, Proc. of he 28h Conf. on Learning Theory (COLT 2015), July 3-6, Paris, France, pages , H. B. McMahan and J. Abernehy. Minimax opimal algorihms for unconsrained linear opimizaion. In C. J. C. Burges, L. Boou, M. Welling, Z. Ghahramani, and K. Q. Weinberger, ediors, Advances in Neural Informaion Processing Sysems 26 (NIPS 2013), pages , H. B. McMahan and F. Orabona. Unconsrained online linear learning in Hilber spaces: Minimax algorihms and normal approximaions. In Vialy Feldman, Csaba Szepesvri, and Maria Florina Balcan, ediors, Proc. of The 27h Conf. on Learning Theory (COLT 2014), June 13-15, Barcelona, Spain, pages , F. Orabona. Dimension-free exponeniaed gradien. In C. J. C. Burges, L. Boou, M. Welling, Z. Ghahramani, and K. Q. Weinberger, ediors, Advances in Neural Informaion Processing Sysems 26, pages Curran Associaes, Inc., F. Orabona. Simulaneous model selecion and opimizaion hrough parameer-free sochasic learning. In Advances in Neural Informaion Processing Sysems 27, pages , F. Orabona and D. Pál. From coin being o parameer-free online learning, Available from: hp: //arxiv.org/pdf/ pdf. S. Shalev-Shwarz. Online learning and online convex opimizaion. Foundaions and Trends in Machine Learning, 4(2): , K. Sridharan, S. Shalev-Shwarz, and N. Srebro. Fas raes for regularized objecives. In D. Koller, D. Schuurmans, Y. Bengio, and L. Boou, ediors, Advances in Neural Informaion Processing Sysems 21, pages Curran Associaes, Inc., M. Sreeer and B. McMahan. No-regre algorihms for unconsrained online convex opimizaion. In F. Pereira, C. J. C. Burges, L. Boou, and K. Q. Weinberger, ediors, Advances in Neural Informaion Processing Sysems 25 (NIPS 2012), Decmber 3-8, Lake Tahoe, Nevada, USA, pages ,

E0 370 Statistical Learning Theory Lecture 20 (Nov 17, 2011)

E0 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 information

An Online Learning-based Framework for Tracking

An 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 information

ANALYSIS AND COMPARISONS OF SOME SOLUTION CONCEPTS FOR STOCHASTIC PROGRAMMING PROBLEMS

ANALYSIS 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 information

A Note on Using the Svensson procedure to estimate the risk free rate in corporate valuation

A Note on Using the Svensson procedure to estimate the risk free rate in corporate valuation A Noe on Using he Svensson procedure o esimae he risk free rae in corporae valuaion By Sven Arnold, Alexander Lahmann and Bernhard Schwezler Ocober 2011 1. The risk free ineres rae in corporae valuaion

More information

11/6/2013. Chapter 14: Dynamic AD-AS. Introduction. Introduction. Keeping track of time. The model s elements

11/6/2013. 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 information

Accelerated Gradient Methods for Stochastic Optimization and Online Learning

Accelerated Gradient Methods for Stochastic Optimization and Online Learning Acceleraed Gradien Mehods for Sochasic Opimizaion and Online Learning Chonghai Hu, James T. Kwok, Weike Pan Deparmen of Compuer Science and Engineering Hong Kong Universiy of Science and Technology Clear

More information

Multiobjective Prediction with Expert Advice

Multiobjective Prediction with Expert Advice Muliobjecive Predicion wih Exper Advice Alexey Chernov Compuer Learning Research Cenre and Deparmen of Compuer Science Royal Holloway Universiy of London GTP Workshop, June 2010 Alexey Chernov (RHUL) Muliobjecive

More information

The Transport Equation

The Transport Equation The Transpor Equaion Consider a fluid, flowing wih velociy, V, in a hin sraigh ube whose cross secion will be denoed by A. Suppose he fluid conains a conaminan whose concenraion a posiion a ime will be

More information

MTH6121 Introduction to Mathematical Finance Lesson 5

MTH6121 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 information

Chapter 7. Response of First-Order RL and RC Circuits

Chapter 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 information

Mathematics in Pharmacokinetics What and Why (A second attempt to make it clearer)

Mathematics 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 information

Chapter 8: Regression with Lagged Explanatory Variables

Chapter 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 information

Cointegration: The Engle and Granger approach

Cointegration: The Engle and Granger approach Coinegraion: The Engle and Granger approach Inroducion Generally one would find mos of he economic variables o be non-saionary I(1) variables. Hence, any equilibrium heories ha involve hese variables require

More information

Real-time Particle Filters

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 information

CHARGE AND DISCHARGE OF A CAPACITOR

CHARGE 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 information

Task is a schedulable entity, i.e., a thread

Task is a schedulable entity, i.e., a thread Real-Time Scheduling Sysem Model Task is a schedulable eniy, i.e., a hread Time consrains of periodic ask T: - s: saring poin - e: processing ime of T - d: deadline of T - p: period of T Periodic ask T

More information

On Stochastic and Worst-case Models for Investing

On Stochastic and Worst-case Models for Investing On Sochasic and Wors-case Models for Invesing Elad Hazan IBM Almaden Research Cener 650 Harry Rd, San Jose, CA 9520 ehazan@cs.princeon.edu Sayen Kale Yahoo! Research 430 Grea America Parkway, Sana Clara,

More information

Random 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 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 information

cooking trajectory boiling water B (t) microwave 0 2 4 6 8 101214161820 time t (mins)

cooking trajectory boiling water B (t) microwave 0 2 4 6 8 101214161820 time t (mins) Alligaor egg wih calculus We have a large alligaor egg jus ou of he fridge (1 ) which we need o hea o 9. Now here are wo accepable mehods for heaing alligaor eggs, one is o immerse hem in boiling waer

More information

On Learning Algorithms for Nash Equilibria

On Learning Algorithms for Nash Equilibria On Learning Algorihms for Nash Equilibria Consaninos Daskalakis 1, Rafael Frongillo 2, Chrisos H. Papadimiriou 2 George Pierrakos 2, and Gregory Valian 2 1 MIT, cosis@csail.mi.edu, 2 UC Berkeley, {raf

More information

adaptive control; stochastic systems; certainty equivalence principle; long-term

adaptive 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 information

Stochastic Optimal Control Problem for Life Insurance

Stochastic 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 information

Optimal Stock Selling/Buying Strategy with reference to the Ultimate Average

Optimal Stock Selling/Buying Strategy with reference to the Ultimate Average Opimal Sock Selling/Buying Sraegy wih reference o he Ulimae Average Min Dai Dep of Mah, Naional Universiy of Singapore, Singapore Yifei Zhong Dep of Mah, Naional Universiy of Singapore, Singapore July

More information

An Online Portfolio Selection Algorithm with Regret Logarithmic in Price Variation

An Online Portfolio Selection Algorithm with Regret Logarithmic in Price Variation An Online Porfolio Selecion Algorihm wih Regre Logarihmic in Price Variaion Elad Hazan echnion - Israel Ins. of ech. Haifa, 32000 Israel ehazan@ie.echnion.ac.il Sayen Kale IBM.J. Wason Research Cener P.O.

More information

Algorithms for Portfolio Management based on the Newton Method

Algorithms for Portfolio Management based on the Newton Method Algorihms for Porfolio Managemen based on he Newon Mehod Ami Agarwal aagarwal@cs.princeon.edu Elad Hazan ehazan@cs.princeon.edu Sayen Kale sayen@cs.princeon.edu Rober E. Schapire schapire@cs.princeon.edu

More information

Measuring macroeconomic volatility Applications to export revenue data, 1970-2005

Measuring 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 information

Journal Of Business & Economics Research September 2005 Volume 3, Number 9

Journal 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 information

DYNAMIC MODELS FOR VALUATION OF WRONGFUL DEATH PAYMENTS

DYNAMIC 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 information

A Distributed Multiple-Target Identity Management Algorithm in Sensor Networks

A 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 information

Making a Faster Cryptanalytic Time-Memory Trade-Off

Making 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 information

A Probability Density Function for Google s stocks

A 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 information

Large Scale Online Learning.

Large 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 information

SPEC model selection algorithm for ARCH models: an options pricing evaluation framework

SPEC 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 information

Permutations and Combinations

Permutations and Combinations Permuaions and Combinaions Combinaorics Copyrigh Sandards 006, Tes - ANSWERS Barry Mabillard. 0 www.mah0s.com 1. Deermine he middle erm in he expansion of ( a b) To ge he k-value for he middle erm, divide

More information

4 Convolution. Recommended Problems. x2[n] 1 2[n]

4 Convolution. Recommended Problems. x2[n] 1 2[n] 4 Convoluion Recommended Problems P4.1 This problem is a simple example of he use of superposiion. Suppose ha a discree-ime linear sysem has oupus y[n] for he given inpus x[n] as shown in Figure P4.1-1.

More information

AP Calculus BC 2010 Scoring Guidelines

AP 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 information

Inventory Planning with Forecast Updates: Approximate Solutions and Cost Error Bounds

Inventory 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 information

Particle Filtering for Geometric Active Contours with Application to Tracking Moving and Deforming Objects

Particle 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 information

Novelty and Collective Attention

Novelty and Collective Attention ovely and Collecive Aenion Fang Wu and Bernardo A. Huberman Informaion Dynamics Laboraory HP Labs Palo Alo, CA 9434 Absrac The subjec of collecive aenion is cenral o an informaion age where millions of

More information

Cost-Sensitive Learning by Cost-Proportionate Example Weighting

Cost-Sensitive Learning by Cost-Proportionate Example Weighting Cos-Sensiive Learning by Cos-Proporionae Example Weighing Bianca Zadrozny, John Langford, Naoki Abe Mahemaical Sciences Deparmen IBM T. J. Wason Research Cener Yorkown Heighs, NY 0598 Absrac We propose

More information

Multiprocessor Systems-on-Chips

Multiprocessor 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 information

Why Did the Demand for Cash Decrease Recently in Korea?

Why 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 information

How To Predict A Person'S Behavior

How 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 information

Biology at Home - Pariion Funcion Guillaume

Biology 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 information

A Short Introduction to Boosting

A Short Introduction to Boosting Journal of Japanese Sociey for Arificial Inelligence,14(5):771-780, Sepember, 1999. (In Japanese, ranslaion by Naoki Abe.) A Shor Inroducion o Boosing Yoav Freund Rober E. Schapire AT&T Labs Research Shannon

More information

9. Capacitor and Resistor Circuits

9. Capacitor and Resistor Circuits ElecronicsLab9.nb 1 9. Capacior and Resisor Circuis Inroducion hus far we have consider resisors in various combinaions wih a power supply or baery which provide a consan volage source or direc curren

More information

Follow the Leader If You Can, Hedge If You Must

Follow the Leader If You Can, Hedge If You Must Journal of Machine Learning Research 15 (2014) 1281-1316 Submied 1/13; Revised 1/14; Published 4/14 Follow he Leader If You Can, Hedge If You Mus Seven de Rooij seven.de.rooij@gmail.com VU Universiy and

More information

The option pricing framework

The option pricing framework Chaper 2 The opion pricing framework The opion markes based on swap raes or he LIBOR have become he larges fixed income markes, and caps (floors) and swapions are he mos imporan derivaives wihin hese markes.

More information

Bayesian Filtering with Online Gaussian Process Latent Variable Models

Bayesian 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 information

Vector Autoregressions (VARs): Operational Perspectives

Vector Autoregressions (VARs): Operational Perspectives Vecor Auoregressions (VARs): Operaional Perspecives Primary Source: Sock, James H., and Mark W. Wason, Vecor Auoregressions, Journal of Economic Perspecives, Vol. 15 No. 4 (Fall 2001), 101-115. Macroeconomericians

More information

Online Multi-Class LPBoost

Online Multi-Class LPBoost Online Muli-Class LPBoos Amir Saffari Marin Godec Thomas Pock Chrisian Leisner Hors Bischof Insiue for Compuer Graphics and Vision, Graz Universiy of Technology, Ausria {saffari,godec,pock,leisner,bischof}@icg.ugraz.a

More information

Simultaneous Perturbation Stochastic Approximation in Decentralized Load Balancing Problem

Simultaneous Perturbation Stochastic Approximation in Decentralized Load Balancing Problem Preprins, 1s IFAC Conference on Modelling, Idenificaion and Conrol of Nonlinear Sysems June 24-26, 2015. Sain Peersburg, Russia Simulaneous Perurbaion Sochasic Approximaion in Decenralized Load Balancing

More information

A Universal Pricing Framework for Guaranteed Minimum Benefits in Variable Annuities *

A Universal Pricing Framework for Guaranteed Minimum Benefits in Variable Annuities * A Universal Pricing Framework for Guaraneed Minimum Benefis in Variable Annuiies * Daniel Bauer Deparmen of Risk Managemen and Insurance, Georgia Sae Universiy 35 Broad Sree, Alana, GA 333, USA Phone:

More information

Duration 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.

Duration 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 information

INTEREST RATE FUTURES AND THEIR OPTIONS: SOME PRICING APPROACHES

INTEREST 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 information

Stability. Coefficients may change over time. Evolution of the economy Policy changes

Stability. Coefficients may change over time. Evolution of the economy Policy changes Sabiliy Coefficiens may change over ime Evoluion of he economy Policy changes Time Varying Parameers y = α + x β + Coefficiens depend on he ime period If he coefficiens vary randomly and are unpredicable,

More information

Online Convex Programming and Generalized Infinitesimal Gradient Ascent

Online Convex Programming and Generalized Infinitesimal Gradient Ascent Online Convex Programming and Generalized Infiniesimal Gradien Ascen Marin Zinkevich Carnegie Mellon Universiy, 5000 Forbes Avenue, Pisburgh, PA 1513 USA maz@cs.cmu.edu Absrac Convex programming involves

More information

Statistical Analysis with Little s Law. Supplementary Material: More on the Call Center Data. by Song-Hee Kim and Ward Whitt

Statistical 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 information

Predicting Stock Market Index Trading Signals Using Neural Networks

Predicting 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 information

Online Learning with Sample Path Constraints

Online Learning with Sample Path Constraints Journal of Machine Learning Research 0 (2009) 569-590 Submied 7/08; Revised /09; Published 3/09 Online Learning wih Sample Pah Consrains Shie Mannor Deparmen of Elecrical and Compuer Engineering McGill

More information

Morningstar Investor Return

Morningstar 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 information

Distributed and Secure Computation of Convex Programs over a Network of Connected Processors

Distributed and Secure Computation of Convex Programs over a Network of Connected Processors DCDIS CONFERENCE GUELPH, ONTARIO, CANADA, JULY 2005 1 Disribued and Secure Compuaion of Convex Programs over a Newor of Conneced Processors Michael J. Neely Universiy of Souhern California hp://www-rcf.usc.edu/

More information

Principal components of stock market dynamics. Methodology and applications in brief (to be updated ) Andrei Bouzaev, bouzaev@ya.

Principal 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 information

Appendix A: Area. 1 Find the radius of a circle that has circumference 12 inches.

Appendix A: Area. 1 Find the radius of a circle that has circumference 12 inches. Appendi A: Area worked-ou s o Odd-Numbered Eercises Do no read hese worked-ou s before aemping o do he eercises ourself. Oherwise ou ma mimic he echniques shown here wihou undersanding he ideas. Bes wa

More information

4. International Parity Conditions

4. International Parity Conditions 4. Inernaional ariy ondiions 4.1 urchasing ower ariy he urchasing ower ariy ( heory is one of he early heories of exchange rae deerminaion. his heory is based on he concep ha he demand for a counry's currency

More information

INSTRUMENTS OF MONETARY POLICY*

INSTRUMENTS OF MONETARY POLICY* Aricles INSTRUMENTS OF MONETARY POLICY* Bernardino Adão** Isabel Correia** Pedro Teles**. INTRODUCTION A classic quesion in moneary economics is wheher he ineres rae or he money supply is he beer insrumen

More information

AP Calculus AB 2013 Scoring Guidelines

AP Calculus AB 2013 Scoring Guidelines AP Calculus AB 1 Scoring Guidelines The College Board The College Board is a mission-driven no-for-profi organizaion ha connecs sudens o college success and opporuniy. Founded in 19, he College Board was

More information

TEMPORAL 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 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 information

Niche Market or Mass Market?

Niche Market or Mass Market? Niche Marke or Mass Marke? Maxim Ivanov y McMaser Universiy July 2009 Absrac The de niion of a niche or a mass marke is based on he ranking of wo variables: he monopoly price and he produc mean value.

More information

Real Time Bid Optimization with Smooth Budget Delivery in Online Advertising

Real 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 information

Network Discovery: An Estimation Based Approach

Network Discovery: An Estimation Based Approach Nework Discovery: An Esimaion Based Approach Girish Chowdhary, Magnus Egersed, and Eric N. Johnson Absrac We consider he unaddressed problem of nework discovery, in which, an agen aemps o formulae an esimae

More information

Stock Trading with Recurrent Reinforcement Learning (RRL) CS229 Application Project Gabriel Molina, SUID 5055783

Stock Trading with Recurrent Reinforcement Learning (RRL) CS229 Application Project Gabriel Molina, SUID 5055783 Sock raing wih Recurren Reinforcemen Learning (RRL) CS9 Applicaion Projec Gabriel Molina, SUID 555783 I. INRODUCION One relaively new approach o financial raing is o use machine learning algorihms o preic

More information

Working Paper No. 482. Net Intergenerational Transfers from an Increase in Social Security Benefits

Working Paper No. 482. Net Intergenerational Transfers from an Increase in Social Security Benefits Working Paper No. 482 Ne Inergeneraional Transfers from an Increase in Social Securiy Benefis By Li Gan Texas A&M and NBER Guan Gong Shanghai Universiy of Finance and Economics Michael Hurd RAND Corporaion

More information

Sampling Time-Based Sliding Windows in Bounded Space

Sampling Time-Based Sliding Windows in Bounded Space Sampling Time-Based Sliding Windows in Bounded Space Rainer Gemulla Technische Universiä Dresden 01062 Dresden, Germany gemulla@inf.u-dresden.de Wolfgang Lehner Technische Universiä Dresden 01062 Dresden,

More information

Chapter 4: Exponential and Logarithmic Functions

Chapter 4: Exponential and Logarithmic Functions Chaper 4: Eponenial and Logarihmic Funcions Secion 4.1 Eponenial Funcions... 15 Secion 4. Graphs of Eponenial Funcions... 3 Secion 4.3 Logarihmic Funcions... 4 Secion 4.4 Logarihmic Properies... 53 Secion

More information

The naive method discussed in Lecture 1 uses the most recent observations to forecast future values. That is, Y ˆ t + 1

The 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 information

Single-machine Scheduling with Periodic Maintenance and both Preemptive and. Non-preemptive jobs in Remanufacturing System 1

Single-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 information

Differential Equations. Solving for Impulse Response. Linear systems are often described using differential equations.

Differential Equations. Solving for Impulse Response. Linear systems are often described using differential equations. Differenial Equaions Linear sysems are ofen described using differenial equaions. For example: d 2 y d 2 + 5dy + 6y f() d where f() is he inpu o he sysem and y() is he oupu. We know how o solve for y given

More information

Strategic Optimization of a Transportation Distribution Network

Strategic 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 information

Performance Center Overview. Performance Center Overview 1

Performance 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 information

17 Laplace transform. Solving linear ODE with piecewise continuous right hand sides

17 Laplace transform. Solving linear ODE with piecewise continuous right hand sides 7 Laplace ransform. Solving linear ODE wih piecewise coninuous righ hand sides In his lecure I will show how o apply he Laplace ransform o he ODE Ly = f wih piecewise coninuous f. Definiion. A funcion

More information

THE PRESSURE DERIVATIVE

THE PRESSURE DERIVATIVE Tom Aage Jelmer NTNU Dearmen of Peroleum Engineering and Alied Geohysics THE PRESSURE DERIVATIVE The ressure derivaive has imoran diagnosic roeries. I is also imoran for making ye curve analysis more reliable.

More information

Nikkei Stock Average Volatility Index Real-time Version Index Guidebook

Nikkei Stock Average Volatility Index Real-time Version Index Guidebook Nikkei Sock Average Volailiy Index Real-ime Version Index Guidebook Nikkei Inc. Wih he modificaion of he mehodology of he Nikkei Sock Average Volailiy Index as Nikkei Inc. (Nikkei) sars calculaing and

More information

Modeling a distribution of mortgage credit losses Petr Gapko 1, Martin Šmíd 2

Modeling a distribution of mortgage credit losses Petr Gapko 1, Martin Šmíd 2 Modeling a disribuion of morgage credi losses Per Gapko 1, Marin Šmíd 2 1 Inroducion Absrac. One of he bigges risks arising from financial operaions is he risk of counerpary defaul, commonly known as a

More information

Chapter 2 Kinematics in One Dimension

Chapter 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 information

Online Max-Margin Weight Learning for Markov Logic Networks

Online Max-Margin Weight Learning for Markov Logic Networks Online Max-Margin Weigh Learning for Markov Logic Neworks Tuyen N. Huynh Raymond J. Mooney Absrac Mos of he exising weigh-learning algorihms for Markov Logic Neworks (MLNs) use bach raining which becomes

More information

Working Paper On the timing option in a futures contract. SSE/EFI Working Paper Series in Economics and Finance, No. 619

Working Paper On the timing option in a futures contract. SSE/EFI Working Paper Series in Economics and Finance, No. 619 econsor www.econsor.eu Der Open-Access-Publikaionsserver der ZBW Leibniz-Informaionszenrum Wirschaf The Open Access Publicaion Server of he ZBW Leibniz Informaion Cenre for Economics Biagini, Francesca;

More information

LIFE INSURANCE WITH STOCHASTIC INTEREST RATE. L. Noviyanti a, M. Syamsuddin b

LIFE INSURANCE WITH STOCHASTIC INTEREST RATE. L. Noviyanti a, M. Syamsuddin b LIFE ISURACE WITH STOCHASTIC ITEREST RATE L. oviyani a, M. Syamsuddin b a Deparmen of Saisics, Universias Padjadjaran, Bandung, Indonesia b Deparmen of Mahemaics, Insiu Teknologi Bandung, Indonesia Absrac.

More information

Dynamic programming models and algorithms for the mutual fund cash balance problem

Dynamic 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 information

Supplementary Appendix for Depression Babies: Do Macroeconomic Experiences Affect Risk-Taking?

Supplementary 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 information

Optimal Investment and Consumption Decision of Family with Life Insurance

Optimal 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 information

Distributing Human Resources among Software Development Projects 1

Distributing 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 information

Evolutionary building of stock trading experts in real-time systems

Evolutionary building of stock trading experts in real-time systems Evoluionary building of sock rading expers in real-ime sysems Jerzy J. Korczak Universié Louis Paseur Srasbourg, France Email: jjk@dp-info.u-srasbg.fr Absrac: This paper addresses he problem of consrucing

More information

EFFICIENT STOCHASTIC MODELING FOR LARGE AND CONSOLIDATED INSURANCE BUSINESS: INTEREST RATE SAMPLING ALGORITHMS

EFFICIENT STOCHASTIC MODELING FOR LARGE AND CONSOLIDATED INSURANCE BUSINESS: INTEREST RATE SAMPLING ALGORITHMS EFFICIENT STOCHASTIC MODELING FOR LARGE AND CONSOLIDATED INSURANCE BUSINESS: INTEREST RATE SAMPLING ALGORITHMS Yvonne C. M. Chueh* Despie he availabiliy of faser, more powerful compuers, he complexiy of

More information

A Bayesian framework with auxiliary particle filter for GMTI based ground vehicle tracking aided by domain knowledge

A 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 information

Verification Theorems for Models of Optimal Consumption and Investment with Retirement and Constrained Borrowing

Verification Theorems for Models of Optimal Consumption and Investment with Retirement and Constrained Borrowing MATHEMATICS OF OPERATIONS RESEARCH Vol. 36, No. 4, November 2, pp. 62 635 issn 364-765X eissn 526-547 364 62 hp://dx.doi.org/.287/moor..57 2 INFORMS Verificaion Theorems for Models of Opimal Consumpion

More information

MULTI-PERIOD OPTIMIZATION MODEL FOR A HOUSEHOLD, AND OPTIMAL INSURANCE DESIGN

MULTI-PERIOD OPTIMIZATION MODEL FOR A HOUSEHOLD, AND OPTIMAL INSURANCE DESIGN Journal of he Operaions Research Sociey of Japan 27, Vol. 5, No. 4, 463-487 MULTI-PERIOD OPTIMIZATION MODEL FOR A HOUSEHOLD, AND OPTIMAL INSURANCE DESIGN Norio Hibiki Keio Universiy (Received Ocober 17,

More information

Economics Honors Exam 2008 Solutions Question 5

Economics Honors Exam 2008 Solutions Question 5 Economics Honors Exam 2008 Soluions Quesion 5 (a) (2 poins) Oupu can be decomposed as Y = C + I + G. And we can solve for i by subsiuing in equaions given in he quesion, Y = C + I + G = c 0 + c Y D + I

More information

Fakultet for informasjonsteknologi, Institutt for matematiske fag

Fakultet for informasjonsteknologi, Institutt for matematiske fag Page 1 of 5 NTNU Noregs eknisk-naurviskaplege universie Fakule for informasjonseknologi, maemaikk og elekroeknikk Insiu for maemaiske fag - English Conac during exam: John Tyssedal 73593534/41645376 Exam

More information