DQL: A New Updating Strategy for Reinforcement Learning Based on Q-Learning

Size: px
Start display at page:

Download "DQL: A New Updating Strategy for Reinforcement Learning Based on Q-Learning"

Transcription

1 DQL: A New Updting Strtegy for Reinforcement Lerning Bsed on Q-Lerning Crlos E. Mrino 1 nd Edurdo F. Morles 2 1 Instituto Mexicno de Tecnologí del Agu, Pseo Cuhunáhuc 8532, Jiutepec, Morelos, 6255, MEXICO cmrino@tlloc.imt.mx 2 ITESM-Cmpus Cuernvc, Pseo de l Reform 182-A, Col. Loms de Cuernvc, Temixco, Morelos, 62589, MEXICO emorles@cmpus.mor.itesm.mx Abstrct. In reinforcement lerning n utonomous gent lerns n optiml policy while intercting with the environment. In prticulr, in one-step Q-lerning, with ech ction n gent updtes its Q vlues considering immedite rewrds. In this pper new strtegy for updting Q vlues is proposed. The strtegy, implemented in n lgorithm clled DQL, uses set of gents ll serching the sme gol in the sme spce to obtin the sme optiml policy. Ech gent leves trces over copy of the environment (copies of Q-vlues), while serching for gol. These copies re used by the gents to decide which ctions to tke. Once ll the gents rech gol, the originl Q-vlues of the best solution found by ll the gents re updted using Wtkins Q-lerning formul. DQL hs some similrities with Gmbrdell s Ant-Q lgorithm [4], however it does not require the definition of domin dependent heuristic nd consequently the tuning of dditionl prmeters. DQL lso does not updte the originl Q-vlues with zero rewrd while the gents re serching, s Ant-Q does. It is shown how DQL s guided explortion of severl gents with selected exploittion (updting only the best solution) produces fster convergence times thn Q-lerning nd Ant-Q on severl testbed problems under similr conditions. 1 Introduction Reinforcement lerning is n on-line technique tht pproximtes dynmic progrmming. The externl environment is modeled s discrete-time, finite stte, Mrkov decision process. Ech ction is ssocited with rewrd. The tsk of reinforcement lerning is to mximize the long-term discounted rewrd per ction. Reinforcement lerning hs been recently pplied to multi gent settings. The min purpose is to coordinte gents to complete tsk. In coordintion problems, ech gent is responsible for portion of the problem, nd most of the time, decisions of n gent ffect other gents performnce or solution. Exmples include the solution of network routing problems in [6] nd coordintion L. De Redt nd P. Flch (Eds.): ECML 21, LNAI 2167, pp , 21. c Springer-Verlg Berlin Heidelberg 21

2 DQL: A New Updting Strtegy for Reinforcement Lerning 325 gmes such s soccer [7]. Multi gent reinforcement lerning hve lso been used in pursuit gmes, where hunter tries to cpture prey. In these problems, gents shre senstions of the loction of the prey, communicte its loction to its prtners nd updte their reltive loction in order rech the prey [13]. Price nd Boutilier [1] proposed method clled implicit imittion. In this pproch pprentice gents lern from the experience of mentor gents bout its own cpbilities in unvisited prts of the spce. Imittion is performed extrcting model from the experienced gent behvior. This pproch ws proved in the solution of mzes using model bsed reinforcement lerning lgorithms, speeding lerning drmticlly. Other interesting problems, solved using multi gent reinforcement lerning, re those known s n-plyer coopertive repeted gmes. In these problems gents interct in limited resource environment selecting ctions tht mximize rewrd. The chosen ctions constitute joint ction. Ech joint ction is ssocited with rewrd function; the decision problem is coopertive since there is single rewrd function reflecting the utility ssessment of ll the gents. Agents must cooperte in order to select those ctions representing the mximl individul nd tem benefit. Some pproches to estblish coopertive behvior between gents for these kind of problems include [1,2,5]. In most of these pproches, single gent reinforcement lerning methods re pplied without much modifiction. In this pper, we propose n lterntive strtegy for updting vlue functions. The min motivtion behind this reserch is to improve the convergence times of Q-lerning with distributed reinforcement lerning setting, where set of gents hve the sme gol, nd together cooperte by leving trces to find n optiml policy for the sme problem. The hypothesis is tht using more explortion with set of gents nd controlled exploittion, by leving trces between gents nd reinforcing only the best solution proposed by the gents, produces fster convergence times. DQL performnce ws compred ginst Q-lerning [14] nd Gmbrdell nd Dorigo s Ant-Q lgorithm [4], which is distributed reinforcement lerning lgorithm used in the solution of the trveling slesmn problem. The three lgorithms were tested on severl problems over the whole rnge of the α nd γ prmeters used in the Q-lerning formul. It is shown tht DQL hs fster convergence times thn one-step Q-lerning nd Ant-Q under similr conditions. The pper is orgnized s follows. Section 2 gives brief overview of Q- lerning nd Ant-Q. Section 3 describes DQL. Section 4, presents the four test problems used to mesure the lgorithms performnce nd discusses the min results. Finlly, Section 5 concludes nd gives future reserch directions. 2 Q-Lerning In this study, ech reinforcement lerning gent uses the one-step Q-lerning lgorithm [14]. Its lerned decision policy is determined by the stte-ction pir vlue function, Q(s, ), which estimtes long-term discounted rewrds for ech stte-ction pir. Given current stte s Snd vilble ctions i A s, Q-lerning gent selects most of the time n ction with the highest estimted

3 326 C.E. Mrino nd E.F. Morles Q(s, ) nd with smll probbility ε, selects n lterntive ction. The gent then executes the ction, receives n immedite rewrd r, nd moves to the next stte s. In ech step, the gent updtes Q(s, ) by recursively discounting future utilities nd weighting them by positive lerning rte α: [ ] Q(s, ) Q(s, )+α r + γ mx Q(s, ) Q(s, ) A s where ( γ 1) is discount prmeter. As n gent explores the stte spce, its estimte Q improves grdully, nd, eventully, ech mx A Q(s, ) pproches: E { } s n=1 γn 1 r t+n. Here rt is the rewrd received t time t due the ction chosen t time t 1. Wtkins nd Dyn [15] hve shown tht this Q-lerning lgorithm converges to n optiml decision policy for finite Mrkov decision process. (1) 2.1 Ant-Q An interesting distributed reinforcement lgorithm, originlly proposed by Gmbrdell nd Dorigo [4], is Ant-Q. Ant-Q ws used to solve trveling slesmn problems nd cn be seen s n improvement over previous system clled nt systems [3]. The generl ide of Ant-Q is to use set of gents serching for the sme best policy. Following n nlogy with nt colonies, ech gent updtes its Q vlues, s in Q-lerning, but without considering ny rewrd (r = in Eq. 1), fter executing ech ction. This updting represent trces tht cn be followed by other gents. Once ll the gents rech gol (n episode), stte-ction pir evlution functions of the best solution re updted using delyed rewrd (r ) s expressed in Eq. 1. This mens tht some Q vlues will be updted severl times on ech episode, first without rewrds by ll the gents tht followed the sme stte-ction pir, nd once more with rewrds if the stte-ction pir is prt of the best pth. This repeted updting is not clerly justified nd is difficult to prove if the convergence properties of Q-lerning still hold. Ant-Q introduced severl dditionl mechnisms to the Q-lerning frmework. In prticulr, the selection policy is defined s combintion of domin dependent heuristic function (HE(s, )) nd the best Q-vlues. This combintion introduces two new prmeters (δ nd β) tht estimte the relevnce of HE(s, ) with respect to Q(s, ) vlues nd tht need to be tuned for ech prticulr ppliction domin. In generl n ɛ-greedy strtegy is used nd HE is combined with Q vlues s follows: rgmx {AQ(s, ) δ HE(s, ) β }. 3 DQL DQL follows similr ides of Ant-Q but without loosing the min properties of Q- lening nor introducing extr prmeters or heuristics. The generl ides, nd min differences with Ant-Q, re tht it does not use ny domin dependent

4 DQL: A New Updting Strtegy for Reinforcement Lerning 327 heuristic (nd consequently no dditionl prmeters) nd it updtes the Q- vlues only once (for the best solution found by ll the gents). DQL llows more explortion, s severl gents re serching t the sme time, nd promotes better exploittion, since the updtes on the Q-vlues re performed only over the best solutions 1. All the gents hve ccess to temporry copy of the stte-ction pir evlution functions (Q C (s, )). Ech time n gent hs to select n ction, it looks t this copy nd decides, bsed on its informtion, which ction to tke. Once the gent performs the selected ction, it updtes the copy of the stte-ction vlue pir using Eq. 2, where Q C (s, ) represents copy of the originl Q(s, ) pirs. [ ] Q C (s, ) Q C (s, )+α γ mx Q C (s, ) Q C (s, ) (2) A s This is similr to wht Ant-Q does, however in this cse the updtes re performed over copies of the originl Q vlues nd the originl Q-vlues re consequently not ffected t this stge. All the gents re moved one step t time, updting nd shring their common Q C vlues until reching stopping criterion. The gents use the copies of the Q vlues to decide which ctions to tke following n ɛ greedy policy. When ll the gents hve found solution the Q vlue copies re discrded nd the stte-ction pirs considered in the best solution receive rewrd which reinforce their vlues ccording to Eq. 1. This updtes the originl Q-vlues from which new copy is creted for the next cycle. The whole process is repeted until reching termintion criterion (see Tble 1). Tble 1. DQL lgorithm. Initilize Q(s, ) rbitrrily Repet (for n episodes) Initilize s, copy Q(s, ) toq C(s, ) Repet (for ech step of episode) Repet (for m gents) Tke ction, observe r, s Q C(s, ) Q C(s, )+α [γ mx Q C(s, ) Q C(s, )] s s ; Until s is terminl Evlute the m proposed solutions Assign rewrds to the best solution found nd updte the Q vlues: Q(s, ) Q(s, )+α [r + γ mx Q(s, ) Q(s, )] 1 In the tested problems, the best solution of one episode is the shortest pth found by one gent in tht episode.

5 328 C.E. Mrino nd E.F. Morles All the gents ct on the sme environment nd hve ccess to the sme Q nd Q C vlues. The copies of the Q vlues re used s guidnces to the gents of wht seems to be promising sttes. However, only the best solution found by ll the gents receives n ctul rewrd. There re two min differences with respect to Ant-Q: Prtil updtes re performed over copies of the Q-vlues voiding multiple updtes with nd without rewrds. There is no need to define domin dependent heuristic or to tune extr prmeters s in Ant-Q. The min motivtion behind DQL is tht it llows: More explortion s more gents re used during serch More exploittion s only relevnt (best) solutions re effectively rewrded The hypothesis is tht this lterntive strtegy for updting vlue functions chieves, in generl, fster convergence times thn one-step Q-lerning, regrdless the vlues of α nd γ. To test this hypothesis, we performed severl experiments over four problems with different complexity nd nture, compring Q-lerning, Ant-Q nd DQL performnce. Although, the tests were performed of deterministic stte trnsition domins, our frmework cn lso be pplied to stochstic stte trnsition functions. 4 Experimentl Results All the experiments were performed on the sme mchine nd the lgorithms were similrly coded by the sme uthor 2. Although DQL nd Ant-Q use multiple gents the lgorithms re implemented sequentilly. Two mze problems were first considered s they re problems where Q- lerning normlly shows good performnce. For these problems the lgorithms were tested over ll possible α nd γ vlues with.25 increments. ε =.1 ws considered for the three lgorithms in both mze problems. Ech experiment ws performed thirty times nd we report the men CPU time, men number of episodes, nd men number of steps per episode 3. Algorithm execution stops when the optiml policy (solid lines in Figure 1) is reched in five consecutive episodes. As mentioned erlier, the Ant-Q lgorithm ws designed for the solution of trveling slesmn problems (TSPs). Two TSP instnces previously solved with Ant-Q re lso included in the tests. The sme prmeter vlues nd stopping criteri used with Ant-Q were used for DQL nd Q-lerning. Tbles of results include best solution found, stndrd devition of solutions, the men of ll the solutions, nd the men CPU time to rech the stopping criterion. In this cse, every lgorithm ws executed 15 times over 2 episodes. 2 All lgorithms re coded in C++. 3 The men number refers to the men of ll the solutions found t prticulr episode.

6 DQL: A New Updting Strtegy for Reinforcement Lerning Grid World with Wind The first experiment ws run on the windy grid world shown in Figure 1 left. The objective is to find the optiml pth from S to G considering wind force, which shifts upwrds the resulting stte when moving horizontlly, the strength of which vries from column to column s shown t the bottom of Figure 1. For instnce, moving horizontlly (either left or right) from squre which hs wind force of 1 (indicted t the bottom of Figure 1), cuses the gent to move one squre bove its intended destintion. However, moving verticlly (either down or up) does not produce ny effect 4. Ant-Q nd DQL were both run with 3 gents. Ant-Q ws tested with nd without heuristic. The results re shown in Figures 2 nd 3 without heuristic for Ant-Q. Restrictions S optiml policy G ctions S G ctions optiml policy Fig. 1. Grid world in which horizontl movement is ltered by loction-dependent upwrd wind (left) nd windy world with restrictions (right). Results from the three lgorithms re plotted using three different line types, dshed for Ant-Q, dsh-dot for Q-lerning, nd continuous for DQL. There re four lines for ech lgorithm, one for ech vlue of γ. The + symbol correspond to γ =.25, to γ =.5, to γ =.75, nd to γ =1.. Figure 2 shows the men CPU time required for ech lgorithm to rech the stopping criterion. For Ant-Q nd DQL this time corresponds to the men time required for ll the solutions found t ech episode. As cn be seen from the results, both Ant-Q nd DQL clerly outperformed Q-lerning for ll the tested vlues of α nd γ, significntly reducing the convergence times. Figure 3 shows the men number of episodes nd the men number of steps per episode required for the three lgorithms to rech the stopping criterion. For these two metrics DQL performnce ws the best of the three lgorithms, nd Q-lening ws ble to outperform Ant-Q for α.5. The previously described results re for Ant-Q without using ny heuristic. When Ant-Q ws tested using s heuristic the inverse of the Mnhttn distnce, 4 An gent is not llowed to move outside the borders.

7 33 C.E. Mrino nd E.F. Morles.9 Grid world.8 Q-lerning.7.6 Ant-Q men CPU time DQL Fig. 2. Men CPU time in seconds to rech the optiml solution five consecutive episodes. 2 Grid world 18 Grid world Q-lerning DQL Ant-Q men number of episodes men number of steps episode Ant-Q Q-lerning 4 4 DQL Fig. 3. Men episodes (left) nd steps per episode (right) required to rech the optiml solution. it ws not ble to converge 5 with three different combintions of (δ, β): (1, 1) Q- vlues nd heuristic function eqully importnt, (2, 1) Q-vlues more importnt thn the heuristic function, nd (1, 2) heuristic function more importnt thn Q- vlues. Although, the heuristic used my be resonble for some mze problems, it is cler tht in generl, finding suitble heuristic my be very difficult tsk. We lso decided to test vrint of DQL, clled DQL-2, where ech gent performs complete episode before strting with the next gent. Tht is, performing m episodes in sequence without shring informtion while performing the tsk. Figure 4 compres the men CPU times of DQL (here s DQL-1) 5 Rech the optiml policy five consecutive episodes before reching 5, trnsitions.

8 DQL: A New Updting Strtegy for Reinforcement Lerning 331 ginst this episodic updting pproch (DQL-2). As it cn be pprecited in the figure, shring informtion while performing tsk reduces convergence times. Although not shown in the pper, due to restrictions in spce, similr results were observed in the other problems..7 Grid world.6 g=.25 DQL-1 DQL-2.5 g=.5 men CPU time.4.3 g=.75 g= g=.25 g=.5 g=.75 g= Fig. 4. Men CPU times in seconds between DQL nd n episodic vrint (DQL-2) We lso mesured the verge number of totl updtes of Q vlues (Q C + Q) in DQL ginst the number of Q updtes of Q-lerning (see Figure 5). As cn be seen in the figure, lthough DQL updtes lrger number of totl Q- vlues, it converges fster. We believe tht the extr informtion shred by the gents during the process helps to reduce convergence times. Similr behvior ws observed on the other problems. 15 Q-Lerning vs DQL-1 1 Q(s,) updtes DQL-1 Q-Lerning Episode Number Fig. 5. Averge number of totl Q-vlue updtes for DQL nd Q-lerning.

9 332 C.E. Mrino nd E.F. Morles 4.2 Grid World with Wind nd Trp This problem ws designed to generte more difficult mze. An obstcle blocking the optiml pth is included to the windy grid world, forcing gents to serch for n lterntive route. Figure 1 right shows the mze nd the optiml policy tht gents must find. The sme opertion conditions nd prmeters used in the previous mze were considered. Figures 6 nd 7 show the mesures for the three metrics obtined with the three lgorithms under study. Agin, the figures show only the performnce of Ant-Q without heuristic s it ws not ble to converge with the Mnhttn distnce heuristic. In Figure 6 it cn be observed tht Ant-Q is ble to outperformed DQL men CPU time for some combintions of α nd γ: (α =.25,γ.25), α =.5,γ =.5, nd (α =1,γ =.5, 1.). It shows, however, to be much more dependent on the vlues of these prmeters. On the other hnd, DQL shows more stble behvior in reltion to the number of episodes nd steps per episode required for the gents to stisfy the stopping criterion. 3.5 Restricted Grid world 3 Q-lerning 2.5 men CPU time Ant-Q DQL Fig. 6. Men CPU time in seconds to rech the optiml solution five consecutive episodes. 4.3 Trveling Slesmn Problems Ant-Q ws originlly developed to solve instnces of TSP. For them, the uthors of Ant-Q reported results where Ant-Q outperformed severl lterntive lgorithms. We took two instnces of TSP with the sme settings used in the originl Ant-Q pper [4]. The first problem is the 3 cities symmetric TSP known s Oliver3 proposed in [9], nd the second problem is the 48 cities symmetric TSP known s Ry48p proposed in [11]. Ant-Q prmeters for the pseudo rndom proportionl ction choice rule were the sme used by Gmbrdell, tht is, β =2., δ =1., α =.1, γ =.3, nd HE(i, j) =1/d i,j, being i, j cities nd d i,j the distnce between them. For

10 DQL: A New Updting Strtegy for Reinforcement Lerning Restricted Grid world 3.5 x 14 Restricted Grid world Q-lerning 1 Q-lerning 3 men number of episodes men number of steps episode Ant-Q Ant-Q 2 DQL.5 DQL Fig. 7. Men episodes (left) nd steps per episode (right) required to rech the optiml solution. DQL nd Q-lerning, the sme vlues for the α nd γ prmeters were used. Q(s, ) vlues were initilized to the inverse of the number of cities times the verge length of edges, nd n ε-greedy selection policy, with ε =.1 6. The lgorithms considered 2 steps or trnsitions for Oliver3 problem, nd 6 for ryp48. The number of gents were the sme in DQL nd Ant-Q, 3 for Oliver3 nd 48 for Ry48p. For Q-lerning single gent ws used. The performnce ws evluted repeting ech tril 15 times. We report the verge performnces. The CPU time correspond to the verge running times to rech the best result. Tble 2. Results for the two TSPs. Oliver3 Ry48p Best DQL Men Std. Dev CPU Q-lerning Men Std. Dev CPU Ant-Q Men HE Std. Dev CPU Ant-Q Men no HE Std. Dev CPU This policy is equivlent to consider q =.9 in the originl Ant-Q lgorithm

11 334 C.E. Mrino nd E.F. Morles In the TSP problems ll strtegies found the sme best solution. Although Ant-Q ws specilly tuned for this type of problems it did not show the best performnce. It is lso interesting to note tht Q-lerning show the lowest verge solution for Ry48p. It is however cler from the results, tht the verge convergence times of DQL re much smller thn the other strtegies. In these prticulr cses, the heuristic function dded to Ant-Q ws useful for reducing convergence times nd stndrd devitions (which ws not the cse for the grid world problems). 5 Conclusions nd Future Work This pper introduces new strtegy for updting Q vlues implemented in n lgorithm clled DQL. DQL uses set of gents serching the sme gol in the sme spce. Trces (copies of Q-vlues updted without rewrds) re used to guide the explortion of gents. The originl Q vlues of only the best solution found by ll the gents is updted using the one-step Q-lerning formul. It ws shown how DQL s guided explortion of severl gents with selected exploittion (updting only the best solution) produces fster convergence times thn Q- lerning nd Ant-Q on severl testbed problems under similr conditions. The heuristic nd extr prmeters needed by Ant-Q does not seem to be producing ny benefits. Additionlly, selecting good heuristic cn be difficult tsk. DQL, on the other hnd, does not require extr prmeters nd shows, in generl, better convergence times. DQL updting strtegy is performed only on the best pth mong m solutions. In order to preserve the convergence properties of Q-lerning, we need to show tht ll the Q-vlue stte-ction pirs hve non zero probbility of being updted. This is prt of our future work. Before trying prllel version, which seems nturl extension, we would like to perform more tests nd compre the results ginst different strtegies for updting Q vlues, such s Monte Crlo. It will lso be interesting to run DQL without updting the copies of the Q-vlues to ssess its influence in the results (which is like running Q-lerning severl times without updting nd then updte the best results found so fr). Acknowledgments. Thnks to the nonymous reviewers for their helpful comments on the initil drft of this pper. The first uthor ws supported by IMTA. This reserch ws supported by CONACyT under grnt 33-A. References 1. Boutilier, C.: Sequentil Optimlity nd Coordintion in Multi gent Systems, In Proc. of IJCAI-99, Stockholm, Sweden, Clus, C., Boutilier, C.: The Dynmics of Reinforcement Lerning in Coopertive Multigents Systems, In Proc. of AAAI-97 Multigent Lerning Workshop, pg , Providence, 1997.

12 DQL: A New Updting Strtegy for Reinforcement Lerning Dorigo, M.: Optimiztion, Lerning, nd Nturl Algorithms, PhD thesis, Politecnico d Milno, Itly, Gmbrdell, L., M., Dorigo, M.: Ant-Q: A reinforcement Lerning Approch to the Trveling Slesmn Problem, In Proceedings of the 12th Interntionl Conference on Mchine Lerning, pp , Morgn Kufmnn, Hu, J., Wellmn, M.: Multigent Reinforcement Lerning: Theoreticl Frmework nd n Algorithm, In Proc. 15th Int. Conf. on Mchine Lerning, pp , Morgn Kufmnn, Littmn, M., Boyn, J.: A Distributed Reinforcement Lerning Scheme for Network Routing, In Proc. Int. Workshop on Applictions of Neurl Networks to Telecommunictions, pp , J. Alspector, et l., (eds.), Lwrence Erlbum, Hillsdle, NJ, Littmn, M.: Mrkov Gmes s Frmework for Multigent Reinforcement Lerning, In Proc. 11th Int. Conf. on Mchine Lerning, pp , New Brunswick, NJ, 1994, Morgn Kufmnn. 8. Mrino, C., Morles, E.: A New Distributed Reinforcement Lerning Algorithm for the solution of Multiple Objective Optimiztion Problems, In O. Ciro et l., eds. Lecture Notes in Artificil Intelligence, 1793: , April Oliver, I., Smith, D., Hollnd, J.R.: A study of Permuttion Crossover Opertors on the Trveling Slesmn Problem, In Proc. 2nd Int. Conf. n Genetic Algorithms, pp , J.J. Grefenstette (ed.), Lwrence Erlbum, Hillsdle, NJ, Price, B., Boutilier, C.: Implicit Imittion in Multigent Reinforcement Lerning, In Proc. 16th Int. Conf. on Mchine Lerning, pp., Reinelt, G.: The Trveling Slesmn: Computtionl Solutions for TSP Applictions, Springer Verlg, Berlin, Sutton, R., Brto, A.: Reinforcement Lerning n Introduction, MIT Press, Cmbridge, MA, Tn, M.: Multigent Reinforcement Lerning: Independent vs. Coopertive Agents, In Proc. 1th Int. Conf. on Mchine Lerning, pp , Amherst, MA, Wtkins, C.: Lerning from Delyed Rewrds. PhD thesis, Cmbridge University, Cmbridge, MA, Wtkins, C., Dyn, P.: Q-Lerning, Mchine Lerning, 3: , 1992.

A New Learning Algorithm for the MAXQ Hierarchical Reinforcement Learning Method

A New Learning Algorithm for the MAXQ Hierarchical Reinforcement Learning Method A New Lerning Algorithm for the MAXQ Hierrchicl Reinforcement Lerning Method Frzneh Mirzzdeh 1, Bbk Behsz 2, nd Hmid Beigy 1 1 Deprtment of Computer Engineering, Shrif University of Technology, Tehrn,

More information

Complete Coverage Path Planning of Mobile Robot Based on Dynamic Programming Algorithm Peng Zhou, Zhong-min Wang, Zhen-nan Li, Yang Li

Complete Coverage Path Planning of Mobile Robot Based on Dynamic Programming Algorithm Peng Zhou, Zhong-min Wang, Zhen-nan Li, Yang Li 2nd Interntionl Conference on Electronic & Mechnicl Engineering nd Informtion Technology (EMEIT-212) Complete Coverge Pth Plnning of Mobile Robot Bsed on Dynmic Progrmming Algorithm Peng Zhou, Zhong-min

More information

Lecture 10 Evolutionary Computation: Evolution strategies and genetic programming

Lecture 10 Evolutionary Computation: Evolution strategies and genetic programming Lecture 10 Evolutionry Computtion: Evolution strtegies nd genetic progrmming Evolution strtegies Genetic progrmming Summry Negnevitsky, Person Eduction, 2011 1 Evolution Strtegies Another pproch to simulting

More information

CSEP 573 Artificial Intelligence Winter 2016

CSEP 573 Artificial Intelligence Winter 2016 CSEP 573 Artificil Intelligence Winter 2016 Luke Zettlemoyer Problem Spces nd Serch slides from Dn Klein, Sturt Russell, Andrew Moore, Dn Weld, Pieter Abbeel, Ali Frhdi Outline Agents tht Pln Ahed Serch

More information

Ball. Player X. Player O. X Goal. O Goal

Ball. Player X. Player O. X Goal. O Goal Generlizing Adversril Reinforcement Lerning Willim T. B. Uther nd Mnuel M. Veloso Computer Science Deprtment Crnegie Mellon University Pittsburgh, PA 15213 futher,velosog@cs.cmu.edu Abstrct Reinforcement

More information

CSCI 446: Artificial Intelligence

CSCI 446: Artificial Intelligence CSCI 446: Artificil Intelligence Serch Instructor: Michele Vn Dyne [These slides were creted by Dn Klein nd Pieter Abbeel for CS188 Intro to AI t UC Berkeley. All CS188 mterils re vilble t http://i.berkeley.edu.]

More information

Today. Search Problems. Uninformed Search Methods. Depth-First Search Breadth-First Search Uniform-Cost Search

Today. Search Problems. Uninformed Search Methods. Depth-First Search Breadth-First Search Uniform-Cost Search Uninformed Serch [These slides were creted by Dn Klein nd Pieter Abbeel for CS188 Intro to AI t UC Berkeley. All CS188 mterils re vilble t http://i.berkeley.edu.] Tody Serch Problems Uninformed Serch Methods

More information

A REINFORCEMENT LEARNING APPROACH TO SCHEDULING DUAL-ARMED CLUSTER TOOLS WITH TIME VARIATIONS

A REINFORCEMENT LEARNING APPROACH TO SCHEDULING DUAL-ARMED CLUSTER TOOLS WITH TIME VARIATIONS A REINFORCEMENT LEARNING APPROACH TO SCHEDULING DUAL-ARMED CLUSTER TOOLS WITH TIME VARIATIONS Ji-Eun Roh (), Te-Eog Lee (b) (),(b) Deprtment of Industril nd Systems Engineering, Kore Advnced Institute

More information

An Efficient Divide and Conquer Algorithm for Exact Hazard Free Logic Minimization

An Efficient Divide and Conquer Algorithm for Exact Hazard Free Logic Minimization An Efficient Divide nd Conquer Algorithm for Exct Hzrd Free Logic Minimiztion J.W.J.M. Rutten, M.R.C.M. Berkelr, C.A.J. vn Eijk, M.A.J. Kolsteren Eindhoven University of Technology Informtion nd Communiction

More information

A Heuristic Approach for Discovering Reference Models by Mining Process Model Variants

A Heuristic Approach for Discovering Reference Models by Mining Process Model Variants A Heuristic Approch for Discovering Reference Models by Mining Process Model Vrints Chen Li 1, Mnfred Reichert 2, nd Andres Wombcher 3 1 Informtion System Group, University of Twente, The Netherlnds lic@cs.utwente.nl

More information

A Transportation Problem Analysed by a New Ranking Method

A Transportation Problem Analysed by a New Ranking Method (IJIRSE) Interntionl Journl of Innovtive Reserch in Science & Engineering ISSN (Online) 7-07 A Trnsporttion Problem Anlysed by New Rnking Method Dr. A. Shy Sudh P. Chinthiy Associte Professor PG Scholr

More information

In the last lecture, we discussed how valid tokens may be specified by regular expressions.

In the last lecture, we discussed how valid tokens may be specified by regular expressions. LECTURE 5 Scnning SYNTAX ANALYSIS We know from our previous lectures tht the process of verifying the syntx of the progrm is performed in two stges: Scnning: Identifying nd verifying tokens in progrm.

More information

Today. CS 188: Artificial Intelligence Fall Recap: Search. Example: Pancake Problem. Example: Pancake Problem. General Tree Search.

Today. CS 188: Artificial Intelligence Fall Recap: Search. Example: Pancake Problem. Example: Pancake Problem. General Tree Search. CS 88: Artificil Intelligence Fll 00 Lecture : A* Serch 9//00 A* Serch rph Serch Tody Heuristic Design Dn Klein UC Berkeley Multiple slides from Sturt Russell or Andrew Moore Recp: Serch Exmple: Pncke

More information

CS 221: Artificial Intelligence Fall 2011

CS 221: Artificial Intelligence Fall 2011 CS 221: Artificil Intelligence Fll 2011 Lecture 2: Serch (Slides from Dn Klein, with help from Sturt Russell, Andrew Moore, Teg Grenger, Peter Norvig) Problem types! Fully observble, deterministic! single-belief-stte

More information

A Comparison of the Discretization Approach for CST and Discretization Approach for VDM

A Comparison of the Discretization Approach for CST and Discretization Approach for VDM Interntionl Journl of Innovtive Reserch in Advnced Engineering (IJIRAE) Volume1 Issue1 (Mrch 2014) A Comprison of the Discretiztion Approch for CST nd Discretiztion Approch for VDM Omr A. A. Shib Fculty

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd processes. Introducing technology

More information

Engineer To Engineer Note

Engineer To Engineer Note Engineer To Engineer Note EE-169 Technicl Notes on using Anlog Devices' DSP components nd development tools Contct our technicl support by phone: (800) ANALOG-D or e-mil: dsp.support@nlog.com Or visit

More information

Solving Problems by Searching. CS 486/686: Introduction to Artificial Intelligence Winter 2016

Solving Problems by Searching. CS 486/686: Introduction to Artificial Intelligence Winter 2016 Solving Prolems y Serching CS 486/686: Introduction to Artificil Intelligence Winter 2016 1 Introduction Serch ws one of the first topics studied in AI - Newell nd Simon (1961) Generl Prolem Solver Centrl

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd processes. Introducing technology

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd business. Introducing technology

More information

Announcements. CS 188: Artificial Intelligence Fall Recap: Search. Today. Example: Pancake Problem. Example: Pancake Problem

Announcements. CS 188: Artificial Intelligence Fall Recap: Search. Today. Example: Pancake Problem. Example: Pancake Problem Announcements Project : erch It s live! Due 9/. trt erly nd sk questions. It s longer thn most! Need prtner? Come up fter clss or try Pizz ections: cn go to ny, ut hve priority in your own C 88: Artificil

More information

Fig.1. Let a source of monochromatic light be incident on a slit of finite width a, as shown in Fig. 1.

Fig.1. Let a source of monochromatic light be incident on a slit of finite width a, as shown in Fig. 1. Answer on Question #5692, Physics, Optics Stte slient fetures of single slit Frunhofer diffrction pttern. The slit is verticl nd illuminted by point source. Also, obtin n expression for intensity distribution

More information

Policy-contingent state abstraction for

Policy-contingent state abstraction for Policy-contingent stte bstrction for hierrchicl MDPs oelle Pineu nd Geoffrey Gordon School of Computer Science Crnegie Mellon University Pittsburgh PA 15213 jpineuggordon @cs.cmu.edu Abstrct Hierrchiclly

More information

Model-based Policy Gradient Reinforcement Learning

Model-based Policy Gradient Reinforcement Learning 776 Model-bsed Policy Grdient Reinforcement Lerning Xin Wng WANGXI~CS.ORST.EDU Thoms G. Dietterlch TGD~CS.ORST.EDU Deprtment of Computer Science, Oregon Stte University, Derborn Hll 102, Corvllis, OR 97330

More information

Heuristic Search for Identical Payoff Bayesian Games

Heuristic Search for Identical Payoff Bayesian Games Heuristic Serch for Identicl Pyoff Byesin Gmes ABSTRACT Frns A. Oliehoek Informtics Institute University of Amsterdm Amsterdm The Netherlnds F.A.Oliehoek@uv.nl Jilles S. Dibngoye Lvl University Cnd University

More information

4452 Mathematical Modeling Lecture 4: Lagrange Multipliers

4452 Mathematical Modeling Lecture 4: Lagrange Multipliers Mth Modeling Lecture 4: Lgrnge Multipliers Pge 4452 Mthemticl Modeling Lecture 4: Lgrnge Multipliers Lgrnge multipliers re high powered mthemticl technique to find the mximum nd minimum of multidimensionl

More information

Slides for Data Mining by I. H. Witten and E. Frank

Slides for Data Mining by I. H. Witten and E. Frank Slides for Dt Mining y I. H. Witten nd E. Frnk Simplicity first Simple lgorithms often work very well! There re mny kinds of simple structure, eg: One ttriute does ll the work All ttriutes contriute eqully

More information

Announcements. CS 188: Artificial Intelligence Fall Recap: Search. Today. General Tree Search. Uniform Cost. Lecture 3: A* Search 9/4/2007

Announcements. CS 188: Artificial Intelligence Fall Recap: Search. Today. General Tree Search. Uniform Cost. Lecture 3: A* Search 9/4/2007 CS 88: Artificil Intelligence Fll 2007 Lecture : A* Serch 9/4/2007 Dn Klein UC Berkeley Mny slides over the course dpted from either Sturt Russell or Andrew Moore Announcements Sections: New section 06:

More information

Text mining: bag of words representation and beyond it

Text mining: bag of words representation and beyond it Text mining: bg of words representtion nd beyond it Jsmink Dobš Fculty of Orgniztion nd Informtics University of Zgreb 1 Outline Definition of text mining Vector spce model or Bg of words representtion

More information

Computing offsets of freeform curves using quadratic trigonometric splines

Computing offsets of freeform curves using quadratic trigonometric splines Computing offsets of freeform curves using qudrtic trigonometric splines JIULONG GU, JAE-DEUK YUN, YOONG-HO JUNG*, TAE-GYEONG KIM,JEONG-WOON LEE, BONG-JUN KIM School of Mechnicl Engineering Pusn Ntionl

More information

On the Detection of Step Edges in Algorithms Based on Gradient Vector Analysis

On the Detection of Step Edges in Algorithms Based on Gradient Vector Analysis On the Detection of Step Edges in Algorithms Bsed on Grdient Vector Anlysis A. Lrr6, E. Montseny Computer Engineering Dept. Universitt Rovir i Virgili Crreter de Slou sin 43006 Trrgon, Spin Emil: lrre@etse.urv.es

More information

File Manager Quick Reference Guide. June Prepared for the Mayo Clinic Enterprise Kahua Deployment

File Manager Quick Reference Guide. June Prepared for the Mayo Clinic Enterprise Kahua Deployment File Mnger Quick Reference Guide June 2018 Prepred for the Myo Clinic Enterprise Khu Deployment NVIGTION IN FILE MNGER To nvigte in File Mnger, users will mke use of the left pne to nvigte nd further pnes

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd business. Introducing technology

More information

OSPF WEIGHT SETTING OPTIMIZATION FOR SINGLE LINK FAILURES

OSPF WEIGHT SETTING OPTIMIZATION FOR SINGLE LINK FAILURES OSP WEIGH SEIG OPIMIZAIO OR SIGLE LIK AILURES Mohmmed H. Sqlli, Sdiq M. Sit, nd Syed Asdullh Computer Engineering Deprtment King hd University of Petroleum & Minerls Dhhrn 31261, Sudi Arbi {sqlli,sdiq,ssd}@kfupm.edu.s

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd processes. Introducing technology

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd business. Introducing technology

More information

LECT-10, S-1 FP2P08, Javed I.

LECT-10, S-1 FP2P08, Javed I. A Course on Foundtions of Peer-to-Peer Systems & Applictions LECT-10, S-1 CS /799 Foundtion of Peer-to-Peer Applictions & Systems Kent Stte University Dept. of Computer Science www.cs.kent.edu/~jved/clss-p2p08

More information

2 Computing all Intersections of a Set of Segments Line Segment Intersection

2 Computing all Intersections of a Set of Segments Line Segment Intersection 15-451/651: Design & Anlysis of Algorithms Novemer 14, 2016 Lecture #21 Sweep-Line nd Segment Intersection lst chnged: Novemer 8, 2017 1 Preliminries The sweep-line prdigm is very powerful lgorithmic design

More information

COMP 423 lecture 11 Jan. 28, 2008

COMP 423 lecture 11 Jan. 28, 2008 COMP 423 lecture 11 Jn. 28, 2008 Up to now, we hve looked t how some symols in n lphet occur more frequently thn others nd how we cn sve its y using code such tht the codewords for more frequently occuring

More information

Policy-contingent state abstraction for hierarchical MDPs

Policy-contingent state abstraction for hierarchical MDPs Policy-contingent stte bstrction for hierrchicl MDPs Joelle Pineu nd Geoffrey Gordon School of Computer Science Crnegie Mellon University Pittsburgh, PA 15213 jpineu,ggordon@cs.cmu.edu Abstrct Hierrchiclly

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd processes. Introducing technology

More information

1. SEQUENCES INVOLVING EXPONENTIAL GROWTH (GEOMETRIC SEQUENCES)

1. SEQUENCES INVOLVING EXPONENTIAL GROWTH (GEOMETRIC SEQUENCES) Numbers nd Opertions, Algebr, nd Functions 45. SEQUENCES INVOLVING EXPONENTIAL GROWTH (GEOMETRIC SEQUENCES) In sequence of terms involving eponentil growth, which the testing service lso clls geometric

More information

MOMDP solving algorithms comparison for safe path planning problems in urban environments

MOMDP solving algorithms comparison for safe path planning problems in urban environments MOMDP solving lgorithms comprison for sfe pth plnning problems in urbn environments Jen-Alexis Delmer 1 nd Yoko Wtnbe 2 nd Croline P. Crvlho Chnel 3 Abstrct This pper tckles problem of UAV sfe pth plnning

More information

FPGA-Based Implementation of Genetic Algorithm for the Traveling Salesman Problem and Its Industrial Application

FPGA-Based Implementation of Genetic Algorithm for the Traveling Salesman Problem and Its Industrial Application FPGA-Bsed Implementtion of Genetic Algorithm for the Trveling Slesmn Problem nd Its Industril Appliction Ioulii Sklirov nd António B.Ferrri Deprtment of Electronics nd Telecommunictions University of Aveiro,

More information

MA1008. Calculus and Linear Algebra for Engineers. Course Notes for Section B. Stephen Wills. Department of Mathematics. University College Cork

MA1008. Calculus and Linear Algebra for Engineers. Course Notes for Section B. Stephen Wills. Department of Mathematics. University College Cork MA1008 Clculus nd Liner Algebr for Engineers Course Notes for Section B Stephen Wills Deprtment of Mthemtics University College Cork s.wills@ucc.ie http://euclid.ucc.ie/pges/stff/wills/teching/m1008/ma1008.html

More information

Parallel Square and Cube Computations

Parallel Square and Cube Computations Prllel Squre nd Cube Computtions Albert A. Liddicot nd Michel J. Flynn Computer Systems Lbortory, Deprtment of Electricl Engineering Stnford University Gtes Building 5 Serr Mll, Stnford, CA 945, USA liddicot@stnford.edu

More information

MATH 25 CLASS 5 NOTES, SEP

MATH 25 CLASS 5 NOTES, SEP MATH 25 CLASS 5 NOTES, SEP 30 2011 Contents 1. A brief diversion: reltively prime numbers 1 2. Lest common multiples 3 3. Finding ll solutions to x + by = c 4 Quick links to definitions/theorems Euclid

More information

12-B FRACTIONS AND DECIMALS

12-B FRACTIONS AND DECIMALS -B Frctions nd Decimls. () If ll four integers were negtive, their product would be positive, nd so could not equl one of them. If ll four integers were positive, their product would be much greter thn

More information

pdfapilot Server 2 Manual

pdfapilot Server 2 Manual pdfpilot Server 2 Mnul 2011 by clls softwre gmbh Schönhuser Allee 6/7 D 10119 Berlin Germny info@cllssoftwre.com www.cllssoftwre.com Mnul clls pdfpilot Server 2 Pge 2 clls pdfpilot Server 2 Mnul Lst modified:

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd business. Introducing technology

More information

A Formalism for Functionality Preserving System Level Transformations

A Formalism for Functionality Preserving System Level Transformations A Formlism for Functionlity Preserving System Level Trnsformtions Smr Abdi Dniel Gjski Center for Embedded Computer Systems UC Irvine Center for Embedded Computer Systems UC Irvine Irvine, CA 92697 Irvine,

More information

Solving Problems by Searching. CS 486/686: Introduction to Artificial Intelligence

Solving Problems by Searching. CS 486/686: Introduction to Artificial Intelligence Solving Prolems y Serching CS 486/686: Introduction to Artificil Intelligence 1 Introduction Serch ws one of the first topics studied in AI - Newell nd Simon (1961) Generl Prolem Solver Centrl component

More information

USING HOUGH TRANSFORM IN LINE EXTRACTION

USING HOUGH TRANSFORM IN LINE EXTRACTION Stylinidis, Efstrtios USING HOUGH TRANSFORM IN LINE EXTRACTION Efstrtios STYLIANIDIS, Petros PATIAS The Aristotle University of Thessloniki, Deprtment of Cdstre Photogrmmetry nd Crtogrphy Univ. Box 473,

More information

Data sharing in OpenMP

Data sharing in OpenMP Dt shring in OpenMP Polo Burgio polo.burgio@unimore.it Outline Expressing prllelism Understnding prllel threds Memory Dt mngement Dt cluses Synchroniztion Brriers, locks, criticl sections Work prtitioning

More information

Exam #1 for Computer Simulation Spring 2005

Exam #1 for Computer Simulation Spring 2005 Exm # for Computer Simultion Spring 005 >>> SOLUTION

More information

Small Business Networking

Small Business Networking Why network is n essentil productivity tool for ny smll business Effective technology is essentil for smll businesses looking to increse the productivity of their people nd processes. Introducing technology

More information

The Distributed Data Access Schemes in Lambda Grid Networks

The Distributed Data Access Schemes in Lambda Grid Networks The Distributed Dt Access Schemes in Lmbd Grid Networks Ryot Usui, Hiroyuki Miygi, Yutk Arkw, Storu Okmoto, nd Noki Ymnk Grdute School of Science for Open nd Environmentl Systems, Keio University, Jpn

More information

UNIT 11. Query Optimization

UNIT 11. Query Optimization UNIT Query Optimiztion Contents Introduction to Query Optimiztion 2 The Optimiztion Process: An Overview 3 Optimiztion in System R 4 Optimiztion in INGRES 5 Implementing the Join Opertors Wei-Png Yng,

More information

Dynamic Programming. Andreas Klappenecker. [partially based on slides by Prof. Welch] Monday, September 24, 2012

Dynamic Programming. Andreas Klappenecker. [partially based on slides by Prof. Welch] Monday, September 24, 2012 Dynmic Progrmming Andres Klppenecker [prtilly bsed on slides by Prof. Welch] 1 Dynmic Progrmming Optiml substructure An optiml solution to the problem contins within it optiml solutions to subproblems.

More information

P(r)dr = probability of generating a random number in the interval dr near r. For this probability idea to make sense we must have

P(r)dr = probability of generating a random number in the interval dr near r. For this probability idea to make sense we must have Rndom Numers nd Monte Crlo Methods Rndom Numer Methods The integrtion methods discussed so fr ll re sed upon mking polynomil pproximtions to the integrnd. Another clss of numericl methods relies upon using

More information

CS481: Bioinformatics Algorithms

CS481: Bioinformatics Algorithms CS481: Bioinformtics Algorithms Cn Alkn EA509 clkn@cs.ilkent.edu.tr http://www.cs.ilkent.edu.tr/~clkn/teching/cs481/ EXACT STRING MATCHING Fingerprint ide Assume: We cn compute fingerprint f(p) of P in

More information

L. Yaroslavsky. Fundamentals of Digital Image Processing. Course

L. Yaroslavsky. Fundamentals of Digital Image Processing. Course L. Yroslvsky. Fundmentls of Digitl Imge Processing. Course 0555.330 Lecture. Imge enhncement.. Imge enhncement s n imge processing tsk. Clssifiction of imge enhncement methods Imge enhncement is processing

More information

Unit #9 : Definite Integral Properties, Fundamental Theorem of Calculus

Unit #9 : Definite Integral Properties, Fundamental Theorem of Calculus Unit #9 : Definite Integrl Properties, Fundmentl Theorem of Clculus Gols: Identify properties of definite integrls Define odd nd even functions, nd reltionship to integrl vlues Introduce the Fundmentl

More information

Coversheet. Publication metadata

Coversheet. Publication metadata Coversheet This is the ccepted mnuscript (post-print version) of the rticle. Contentwise, the ccepted mnuscript version is identicl to the finl published version, but there my be differences in typogrphy

More information

Epson Projector Content Manager Operation Guide

Epson Projector Content Manager Operation Guide Epson Projector Content Mnger Opertion Guide Contents 2 Introduction to the Epson Projector Content Mnger Softwre 3 Epson Projector Content Mnger Fetures... 4 Setting Up the Softwre for the First Time

More information

What are suffix trees?

What are suffix trees? Suffix Trees 1 Wht re suffix trees? Allow lgorithm designers to store very lrge mount of informtion out strings while still keeping within liner spce Allow users to serch for new strings in the originl

More information

Elena Baralis, Silvia Chiusano Politecnico di Torino. Pag. 1. Query optimization. DBMS Architecture. Query optimizer. Query optimizer.

Elena Baralis, Silvia Chiusano Politecnico di Torino. Pag. 1. Query optimization. DBMS Architecture. Query optimizer. Query optimizer. DBMS Architecture SQL INSTRUCTION OPTIMIZER Dtbse Mngement Systems MANAGEMENT OF ACCESS METHODS BUFFER MANAGER CONCURRENCY CONTROL RELIABILITY MANAGEMENT Index Files Dt Files System Ctlog DATABASE 2 Query

More information

Improved Fast Replanning for Robot Navigation in Unknown Terrain

Improved Fast Replanning for Robot Navigation in Unknown Terrain Improved Fst Replnning for Robot Nvigtion in Unknown Terrin ollege of omputing Georgi Institute of Technology tlnt, G 0-00 GIT-OGSI-00/ Sven Koenig ollege of omputing Georgi Institute of Technology tlnt,

More information

Knowledge States: A Tool in Randomized Online Algorithms

Knowledge States: A Tool in Randomized Online Algorithms : A Tool in Rndomized Online Algorithms Center for the Advnced Study of Algorithms School of Computer Science University of Nevd, Ls Vegs ADS 2007 couthors: Lwrence L. Lrmore, John Nog, Rüdiger Reischuk

More information

this grammar generates the following language: Because this symbol will also be used in a later step, it receives the

this grammar generates the following language: Because this symbol will also be used in a later step, it receives the LR() nlysis Drwcks of LR(). Look-hed symols s eplined efore, concerning LR(), it is possile to consult the net set to determine, in the reduction sttes, for which symols it would e possile to perform reductions.

More information

Efficient Regular Expression Grouping Algorithm Based on Label Propagation Xi Chena, Shuqiao Chenb and Ming Maoc

Efficient Regular Expression Grouping Algorithm Based on Label Propagation Xi Chena, Shuqiao Chenb and Ming Maoc 4th Ntionl Conference on Electricl, Electronics nd Computer Engineering (NCEECE 2015) Efficient Regulr Expression Grouping Algorithm Bsed on Lbel Propgtion Xi Chen, Shuqio Chenb nd Ming Moc Ntionl Digitl

More information

A Probabilistic Emergent Routing Algorithm for Mobile Ad Hoc Networks

A Probabilistic Emergent Routing Algorithm for Mobile Ad Hoc Networks Author mnuscript, published in "WiOpt'03: Modeling nd Optimiztion in Mobile, Ad oc nd Wireless Networks (2003) 10 pges" A Probbilistic Emergent Routing Algorithm for Mobile Ad oc Networks John S. Brs nd

More information

Stack. A list whose end points are pointed by top and bottom

Stack. A list whose end points are pointed by top and bottom 4. Stck Stck A list whose end points re pointed by top nd bottom Insertion nd deletion tke plce t the top (cf: Wht is the difference between Stck nd Arry?) Bottom is constnt, but top grows nd shrinks!

More information

Compilers Spring 2013 PRACTICE Midterm Exam

Compilers Spring 2013 PRACTICE Midterm Exam Compilers Spring 2013 PRACTICE Midterm Exm This is full length prctice midterm exm. If you wnt to tke it t exm pce, give yourself 7 minutes to tke the entire test. Just like the rel exm, ech question hs

More information

ECE 468/573 Midterm 1 September 28, 2012

ECE 468/573 Midterm 1 September 28, 2012 ECE 468/573 Midterm 1 September 28, 2012 Nme:! Purdue emil:! Plese sign the following: I ffirm tht the nswers given on this test re mine nd mine lone. I did not receive help from ny person or mteril (other

More information

1 Introduction

1 Introduction Published in IET Computers & Digitl Techniques Received on 6th July 2006 Revised on 21st September 2007 ISSN 1751-8601 Hrdwre rchitecture for high-speed rel-time dynmic progrmming pplictions B. Mtthews

More information

Approximation by NURBS with free knots

Approximation by NURBS with free knots pproximtion by NURBS with free knots M Rndrinrivony G Brunnett echnicl University of Chemnitz Fculty of Computer Science Computer Grphics nd Visuliztion Strße der Ntionen 6 97 Chemnitz Germny Emil: mhrvo@informtiktu-chemnitzde

More information

SUPPLEMENTARY INFORMATION

SUPPLEMENTARY INFORMATION Supplementry Figure y (m) x (m) prllel perpendiculr Distnce (m) Bird Stndrd devition for distnce (m) c 6 prllel perpendiculr 4 doi:.8/nture99 SUPPLEMENTARY FIGURE Confirmtion tht movement within the flock

More information

2014 Haskell January Test Regular Expressions and Finite Automata

2014 Haskell January Test Regular Expressions and Finite Automata 0 Hskell Jnury Test Regulr Expressions nd Finite Automt This test comprises four prts nd the mximum mrk is 5. Prts I, II nd III re worth 3 of the 5 mrks vilble. The 0 Hskell Progrmming Prize will be wrded

More information

AI Adjacent Fields. This slide deck courtesy of Dan Klein at UC Berkeley

AI Adjacent Fields. This slide deck courtesy of Dan Klein at UC Berkeley AI Adjcent Fields Philosophy: Logic, methods of resoning Mind s physicl system Foundtions of lerning, lnguge, rtionlity Mthemtics Forml representtion nd proof Algorithms, computtion, (un)decidility, (in)trctility

More information

A Scalable and Reliable Mobile Agent Computation Model

A Scalable and Reliable Mobile Agent Computation Model A Sclble nd Relible Mobile Agent Computtion Model Yong Liu, Congfu Xu, Zhohui Wu, nd Yunhe Pn College of Computer Science, Zhejing University Hngzhou 310027, Chin cckffe@yhoo.com.cn Abstrct. This pper

More information

CS321 Languages and Compiler Design I. Winter 2012 Lecture 5

CS321 Languages and Compiler Design I. Winter 2012 Lecture 5 CS321 Lnguges nd Compiler Design I Winter 2012 Lecture 5 1 FINITE AUTOMATA A non-deterministic finite utomton (NFA) consists of: An input lphet Σ, e.g. Σ =,. A set of sttes S, e.g. S = {1, 3, 5, 7, 11,

More information

SOME EXAMPLES OF SUBDIVISION OF SMALL CATEGORIES

SOME EXAMPLES OF SUBDIVISION OF SMALL CATEGORIES SOME EXAMPLES OF SUBDIVISION OF SMALL CATEGORIES MARCELLO DELGADO Abstrct. The purpose of this pper is to build up the bsic conceptul frmework nd underlying motivtions tht will llow us to understnd ctegoricl

More information

Journal of Graph Algorithms and Applications

Journal of Graph Algorithms and Applications Journl of Grph Algorithms nd Applictions http://www.cs.brown.edu/publictions/jg/ vol. 5, no. 5, pp. 17 38 (2001) New Bounds for Oblivious Mesh Routing Kzuo Iwm School of Informtics Kyoto University Kyoto

More information

CSE 401 Midterm Exam 11/5/10 Sample Solution

CSE 401 Midterm Exam 11/5/10 Sample Solution Question 1. egulr expressions (20 points) In the Ad Progrmming lnguge n integer constnt contins one or more digits, but it my lso contin embedded underscores. Any underscores must be preceded nd followed

More information

Modeling and Simulation of Short Range 3D Triangulation-Based Laser Scanning System

Modeling and Simulation of Short Range 3D Triangulation-Based Laser Scanning System Modeling nd Simultion of Short Rnge 3D Tringultion-Bsed Lser Scnning System Theodor Borngiu Anmri Dogr Alexndru Dumitrche April 14, 2008 Abstrct In this pper, simultion environment for short rnge 3D lser

More information

Presentation Martin Randers

Presentation Martin Randers Presenttion Mrtin Rnders Outline Introduction Algorithms Implementtion nd experiments Memory consumption Summry Introduction Introduction Evolution of species cn e modelled in trees Trees consist of nodes

More information

PPS: User Manual. Krishnendu Chatterjee, Martin Chmelik, Raghav Gupta, and Ayush Kanodia

PPS: User Manual. Krishnendu Chatterjee, Martin Chmelik, Raghav Gupta, and Ayush Kanodia PPS: User Mnul Krishnendu Chtterjee, Mrtin Chmelik, Rghv Gupt, nd Ayush Knodi IST Austri (Institute of Science nd Technology Austri), Klosterneuurg, Austri In this section we descrie the tool fetures,

More information

1 Quad-Edge Construction Operators

1 Quad-Edge Construction Operators CS48: Computer Grphics Hndout # Geometric Modeling Originl Hndout #5 Stnford University Tuesdy, 8 December 99 Originl Lecture #5: 9 November 99 Topics: Mnipultions with Qud-Edge Dt Structures Scribe: Mike

More information

Determining Single Connectivity in Directed Graphs

Determining Single Connectivity in Directed Graphs Determining Single Connectivity in Directed Grphs Adm L. Buchsbum 1 Mrtin C. Crlisle 2 Reserch Report CS-TR-390-92 September 1992 Abstrct In this pper, we consider the problem of determining whether or

More information

INTRODUCTION TO SIMPLICIAL COMPLEXES

INTRODUCTION TO SIMPLICIAL COMPLEXES INTRODUCTION TO SIMPLICIAL COMPLEXES CASEY KELLEHER AND ALESSANDRA PANTANO 0.1. Introduction. In this ctivity set we re going to introduce notion from Algebric Topology clled simplicil homology. The min

More information

On String Matching in Chunked Texts

On String Matching in Chunked Texts On String Mtching in Chunked Texts Hnnu Peltol nd Jorm Trhio {hpeltol, trhio}@cs.hut.fi Deprtment of Computer Science nd Engineering Helsinki University of Technology P.O. Box 5400, FI-02015 HUT, Finlnd

More information

CS 321 Programming Languages and Compilers. Bottom Up Parsing

CS 321 Programming Languages and Compilers. Bottom Up Parsing CS 321 Progrmming nguges nd Compilers Bottom Up Prsing Bottom-up Prsing: Shift-reduce prsing Grmmr H: fi ; fi b Input: ;;b hs prse tree ; ; b 2 Dt for Shift-reduce Prser Input string: sequence of tokens

More information

CS 268: IP Multicast Routing

CS 268: IP Multicast Routing Motivtion CS 268: IP Multicst Routing Ion Stoic April 5, 2004 Mny pplictions requires one-to-mny communiction - E.g., video/udio conferencing, news dissemintion, file updtes, etc. Using unicst to replicte

More information

UT1553B BCRT True Dual-port Memory Interface

UT1553B BCRT True Dual-port Memory Interface UTMC APPICATION NOTE UT553B BCRT True Dul-port Memory Interfce INTRODUCTION The UTMC UT553B BCRT is monolithic CMOS integrted circuit tht provides comprehensive MI-STD- 553B Bus Controller nd Remote Terminl

More information

Chapter 2 Sensitivity Analysis: Differential Calculus of Models

Chapter 2 Sensitivity Analysis: Differential Calculus of Models Chpter 2 Sensitivity Anlysis: Differentil Clculus of Models Abstrct Models in remote sensing nd in science nd engineering, in generl re, essentilly, functions of discrete model input prmeters, nd/or functionls

More information

Representation of Numbers. Number Representation. Representation of Numbers. 32-bit Unsigned Integers 3/24/2014. Fixed point Integer Representation

Representation of Numbers. Number Representation. Representation of Numbers. 32-bit Unsigned Integers 3/24/2014. Fixed point Integer Representation Representtion of Numbers Number Representtion Computer represent ll numbers, other thn integers nd some frctions with imprecision. Numbers re stored in some pproximtion which cn be represented by fixed

More information

Dr. D.M. Akbar Hussain

Dr. D.M. Akbar Hussain Dr. D.M. Akr Hussin Lexicl Anlysis. Bsic Ide: Red the source code nd generte tokens, it is similr wht humns will do to red in; just tking on the input nd reking it down in pieces. Ech token is sequence

More information

A dynamic multicast tree based routing scheme without replication in delay tolerant networks

A dynamic multicast tree based routing scheme without replication in delay tolerant networks Accepted Mnuscript A dynmic multicst tree bsed routing scheme without repliction in dely tolernt networks Yunsheng Wng, Jie Wu PII: S0-()00- DOI: 0.0/j.jpdc.0..00 Reference: YJPDC To pper in: J. Prllel

More information

10.5 Graphing Quadratic Functions

10.5 Graphing Quadratic Functions 0.5 Grphing Qudrtic Functions Now tht we cn solve qudrtic equtions, we wnt to lern how to grph the function ssocited with the qudrtic eqution. We cll this the qudrtic function. Grphs of Qudrtic Functions

More information