CACO : Competitive Ant Colony Optimization, A Nature-Inspired Metaheuristic For Large-Scale Global Optimization
aa r X i v : . [ c s . N E ] D ec CACO : Competitive Ant Colony Optimization, ANature-Inspired Metaheuristic For Large-Scale GlobalOptimization
M. A. El-Dosuky a a Computer sciences Department, Faculty of Computers and Information, P.D.Box35516, Mansoura University, Egypt
Abstract
Large-scale problems are nonlinear problems that need metaheuristics, orglobal optimization algorithms. This paper reviews nature-inspired meta-heuristics, then it introduces a framework named Competitive Ant ColonyOptimization inspired by the chemical communications among insects. Thena case study is presented to investigate the proposed framework for large-scaleglobal optimization.
Keywords:
Ant Colony Optimization, Metaheuristics, large-scale global optimization
1. Introduction
Large-scale problems are nonlinear problems that need metaheuristics, orglobal optimization algorithms ([10],[15]).This paper reviews nature-inspired metaheuristics, then it introduces aframework named Competitive Ant Colony Optimization inspired by thechemical communications among insects. Then a case study is presentedto investigate the proposed framework. Finally, the paper concludes with adiscussion of future works.
2. Nature-Inspired Metaheuristics
Large-scale problems are nonlinear problems that need metaheuristics,or global optimization algorithms ([10],[15]). Many nature-inspired meta-heuristic optimization algorithms are proposed to imitate the best behaviors
Preprint submitted to Information Sciences August 20, 2018 n nature [34] such as the artificial immune system [12], genetic algorithms[17], ant colony optimization (ACO) [11], particle swarm optimization (PSO)[22], Artificial Bee Colony Algorithm (ABC) [21], and cuckoo search [35]. Re-cently, many nature inspired algorithms are proposed([67], [68], [69], [70]).A metaheuristic explores the search space by employing two components ofintensification and diversification ([36], [14]). Intensification strategy focuseson examining neighbors of elite solutions while diversification strategy en-courages examining unvisited regions [16]. Intensification is a deterministiccomponent and diversification is a stochastic component [20]. Metaheuristicalgorithms should be designed so that intensification and diversification playbalanced roles [8].
Ant colony optimization (ACO) generates artificial ants that move onthe problem graph depositing artificial pheromone so that the future artificialants can build better solutions ([11], [64]). ACO has been successfully appliedto an impressive number of optimization problems especially for routing andscheduling problems ([65], [66]).ACO proves reliability in large-scale applications such as large-distortedfingerprint matching [63] and solving the logistics problem arising in disasterrelief activities[62].
3. Competitive Ant Colony Optimization
It is probably safe to say that insects rely more heavily on chemical sig-nals than on any other form of communication. These signals, often calledsemiochemicals or infochemicals, serve as a form of language that helps tomediate interactions between organisms. Insects may be highly sensitive tolow concentrations of these chemicals in some cases, a few molecules may beenough to elicit a response. Semiochemicals can be divided into Pheromonesand Allelochemicals based on who sends a message and who receives it [61].Pheromones are chemical signals that carry information from one individ-ual to another member of the same species. These include sex attractants,trail marking compounds, alarm substances, and many other intraspecificmessages. Allelochemicals are signals that travel from one animal to somemember of a different species. These include defensive signals such as re-pellents, compounds used to locate suitable host plants, and a vast array ofother substances that regulate interspecific behaviors.2llelochemicals can be further subdivided into three groups based on whobenefits from the message:
Allomones benefit the sender such as a repellent, or defensive compound(e.g. cyanide) that deters predation..
Kairomones benefit the receiver – such as an odor that a parasite uses tofind its host.
Synomones benefit both sender and receiver – such as plant volatiles thatattract insect pollinators..The diffusion equation of chemical signals is defined as[2]: K = Q Dπr efrec r √ Dt where Q, D, and K are emission rate, diffusion coefficient, and thresholdconcentration, respectively, and where r is the radius of the active space(cm), t is the time from the beginning of emission, and where efrc(x) is thecomplementary error function. In their search for food, ants use pheromones to communicate. Assum-ing there is a natural battle between ants and their enemy that producesallomones. The enemy can be other incest species or ants of different kind.Let assume that the enemy is a group of wasps. Based upon these assump-tions, let us propose the following scenario.First, Ants and Wasps are two groups of insects, competing in the sameenvironment to search for food. Each group behaves like traditional ACOalgorithm. Second, within the same group, communication is done usingpheromones. Communication between the two groups is done using al-lomones. Third, Wasps can kill ants if they are close enough.
Implementation of this modification is done in Microsoft Visual C class Wasp : ACO.ACO{ public Wasp(Dataset data, float evapore, float aging,float limit, bool useOptimize): base(data, evapore, aging, limit, useOptimize){ public override Graph Optimize(params float[]parameters){ return this.ThisGraph;}}class Ant : ACO.ACO{ Wasp wasps;public Ant(Wasp wasps, Dataset data, float evapore,float aging, float limit, bool useOptimize): base(data, evapore, aging, limit, useOptimize){ this.Optimize(1, 1, 1);this.wasps = wasps;}public override Graph Optimize(params float[]parameters){ Graph g = new Graph(this);float r = parameters[0];float d = parameters[1];float q = parameters[2];int k;double sum = 0;int iterations = 1000;for (int t = 1; t < iterations; t++){ sum = q / (2*Math.PI*r) * r / (Math.Sqrt(4*d*t));k =(int) sum * iterations;efrc( new Graph(this.wasps) , this.ThisGraph);} eturn this.ThisGraph;}public void efrc(Graph w, Graph a){ w.Complement(a);}}class Program{ static void Main(string[] args){ Dataset d1 = Dataset.Load(Datasets.Audiology);Dataset d2 = Dataset.Load(Datasets.BreastCancer);Dataset d3 = Dataset.Load(Datasets.Mushroom);Dataset d4 = Dataset.Load(Datasets.Vote);Dataset d5 = Dataset.Load(Datasets.Wine);work(d1);work(d2);work(d3);work(d4);work(d5);Console.ReadKey();}static void work(Dataset dataset){ Wasp wasps = new Wasp(dataset, 0.2F, 1.0F, 1.0F,false);Ant ants = new Ant(wasps, dataset, 0.2F, 1.0F, 1.0F, true);Console.WriteLine(string.Join("\t", new string[]{dataset.name ,dataset.size.ToString(),dataset.feats.ToString(),wasps.feats.ToString(),ants.feats.ToString() }));}
4. Evaluation
Experiments are carried out on five datasets which are all from UCIdatasets ( http://archive.ics.uci.edu/ml/datasets.html ). Inorder to find whether our algorithm could find an optimal reduct, we com-pare algorithm of with traditional method . The experiments are summarizedin Table 1. Table 1: ComparisonDataset Instants Features ACO Proposed CACOAudiology 200 70 20 12Breast Cancer 699 10 4 4Mushroom 8124 23 6 5Wine 178 14 6 5Vote 435 17 12 10
5. Conclusion
This paper reviews Ant Colony optimization algorithms and describes anew heuristic optimization method based on swarm intelligence. It presents amechanism for enhancing Ant colony optimization by introducing the naturalbattle between ants and their enemy that produces Allomones. It is verysimple, easily implemented and it needs fewer parameters, which made itfully developed and applied for feature extraction task
References [1] Ackley, D.H., Connectionist machine for genetic hill climbing, Kluwer,Boston, 1987. 62] G. M. Happ, Chemical signals among animals: Allomones andpheromones, In: Humoral Control of Growth and Differentiation, Non-vertebrate Neuroendocrinology and Aging, Vol. II (J. LoBue and A. S.Gordon, eds.), pp. 149-190. Academic Press,NY. 1974.[3] Arora, J., Introduction to Optimum Design, McGraw-Hill,1989[4] Back, T., Schwefel, H.P., An overview of evolutionary algorithms for pa-rameter optimization, Evolutionary Computation, Vol. 1, No. 1, PP 1-23,1993.[5] Back, T., Michalewiccz, Z., Test landscapes, In Back, T., Fogel, D.B.,Michalewicz, Z., (Ed): Handbook of Evolutionary Computation, ChapterB2.7, PP 14-20, Institute of Physics Publishing and Oxford UniversityPress, New York, 1997.[6] Bartumeus, F. et al. Optimizing the encounter rate in biological inter-actions: Levy versus Brownian strategies. Phys. Rev. Lett. 88, 097901 ,2002.[7] Bartumeus, F., et al., J. Animal search strategies: a quantitative random-walk analysis. Ecology 86, 3078-3087 2005.[8] Blum, C. and Roli, A., ’Metaheuristics in combinatorial optimization:Overview and conceptural comparision’, ACM Comput. Surv., 35, 268-308, 2003[9] De Jong, K. A., An analysis of the behavior of a class of genetic adap-tive systems, Doctoral dissertation, University of Michigan, Ann Arbor,University Microfilms No 76-9381, 1975.[10] Deb, K., Optimisation for Engineering Design, Prentice-Hall, NewDelhi.1995.[11] Dorigo, M,. Optimization, Learning and Natural Algorithms (Phd The-sis). Politecnico di Milano, Italie, 1992[12] Farmer, J.D.; Packard, N.; Perelson, A., The immune system, adapta-tion and machine learning. Physica D 22 (1-3): 187-204. 1986713] Feduccia, Alan, The Bony Stapes in the Upupidae and Phoeniculidae:Evidence for Common Ancestry. The Wilson Bulletin 87 (3): 416-417,1975.[14] Gazi, K., and Passino, K. M., Stability analysis of social foragingswarms, IEEE Trans. Sys. Man. Cyber. Part B - Cybernetics, 34, 539-557,2004.[15] Glover, F. Future Paths for Integer Programming and Links to ArtificialIntelligence. Computers and Operations Research 13 (5): 533-549, 1986[16] Glover, F. and M. Laguna. Tabu Search. Kluwer Academic Publishers,1997.[17] Goldberg, D. E., Genetic Algorithms in Search, Optimisation and Ma-chine Learning, Reading, Mass., Addison Wesley, 1989..[18] Hackett, Shannon J.; et al, ”A Phylogenomic Study of Birds RevealsTheir Evolutionary History”. Science 320 (1763): 1763-1768, 2008[19] Hoffmeister, F., Back. T., Genetic algorithms and Evolution strategies:Similarities and differences, In Shwefel, H-P., Manner. R., Parallel Prob-lem Solving from Nature- PPSN 1 (Lecture Note in Computer Science;Vol. 496), Springer Verlag, Berlin, 1991.[20] Hoos, H. H., T. St¨utzle. Stochastic Local Search. Morgan Kaufmann,2005.[21] Karaboga, D., An Idea Based On Honey Bee Swarm For Numerical Nu-merical Optimization. Technical Report-TR06 (Erciyes University, Engi-neering Faculty, Computer Engineering Department), 2005[22] Kennedy, J.; Eberhart, R. Particle Swarm Optimization. Proceedings ofIEEE International Conference on Neural Networks. IV. pp. 1942-1948,1995[23] Kristin, A. Family Upupidae (Hoopoes). In Josep, del Hoyo; Andrew,Elliott; Sargatal, Jordi. Handbook of the Birds of the World. Volume 6,Mousebirds to Hornbills. Barcelona: Lynx Edicions. pp. 396-411, 2001824] LaGrega, Michael D.; Buckingham, Phillip L.; and Evans, Jeffrey C.Hazardous Waste Management, 2nd edition. New York: McGraw-Hill,2001[25] Pavlyukevich, I.,. Levy flights, non-local search and simulated annealing,J. Computational Physics, 226, 1830-1844, 2007[26] Rajabioun, R., Cuckoo Optimization Algorithm, Applied Soft Comput-ing, Volume 11, Issue 8, Pages 5508-5518, 2011[27] Rosenbrock, H.H., An Automatic method for finding the greatest orleast value of a function, The Computer Journal, Vol. 3, No.3, PP 175-184, 1960.[28] Schoen, F., A wide class of test functions for global optimization, J.Global Optimization, 3, 133-137, 1993[29] Shang, Y. W., Qiu Y. H., A note on the extended rosenrbock function,Evolutionary Computation, 14, 119-126., 2006[30] Torn, A., Zilinskas, A., Global Optimization, Lecture Note in ComputerScience; Vol. 350, Springer Verlarg, Barlin, 1989.[31] Viswanathan, G. M. et al. Optimizing the success of random searches.Nature 401, 911-914 ,1999.[32] Walton, S., Hassan, O., Morgan, K., Brown. M.R., Modified cuckoosearch: A new gradient free optimisation algorithm, Chaos, Solitons andFractals, Volume 44, Issue 9, Pages 710-718, 2011[33] Whitley, L. D., Mathias, K.E., Rana, S., Dzubera, J., Building bettertest functions, In Eshelman, L.J. (Ed), Proceeding of the Sixth Inter-national Conference on Genetic Algorithms, PP 239-246, Morgan Kauf-mann, California, 1995.[34] Yang, X. S., Nature-Inspired Metaheuristic Algorithms, Luniver Press,2008[35] Yang, X.-S.; Deb, S. Cuckoo search via Levy flights, in: World Congresson Nature and Biologically Inspired Computing (NaBIC 2009).. IEEEPublication, USA. pp. 210-214. 2009936] Z¨apfel, G., Roland Braune, Michael B¨ogl, Metaheuristic Search Con-cepts: A Tutorial with Applications to Production and Logistics Books.Springer; 2010[37] Leslie Lamport,
A Document Preparation System . Addison Wesley, Mas-sachusetts, 2nd Edition, 1994.[38] aSusan E. Brennan and Eric A. Hulteen,
Interaction and feedback ina spoken language system: A theoretical framework . Knowledge-BasedSystems, 1995, volume 8, pages 143-151, issue 2.[39] H. Clark and D. Wilkes-Gibbs,
Referring as a collaborative process . Cog-nition, 1986, volume 22, pages 1-39, number 1.[40] Herbert H. Clark,
Using Language , Cambridge University Press, 1996.[41] Herbert H. Clark and Meredyth A. Krych,
Speaking while monitoringaddressees for understanding , Journal of Memory and Language, 2004,volume 50, pages 62-81.[42] Catherine Durnell Cramton,
The Mutual Knowledge Problem and ItsConsequences for Dispersed Collaboration , Organization Science, 2001,volume 12, pages 346-371, number 3.[43] Darren Gergle and Robert E. Kraut and Susan R. Fussell,
LanguageEfficiency and Visual Technology: Minimizing Collaborative Effort withVisual Information , Journal of Language and Social Psychology, 2004,volume 23, pages 491-517, number 4.[44] Ellen A. Issacs and Herbert H. Clark,
References in Conversation Be-tween Experts and Novices , Journal of Experimental Psychology: Gen-eral, 1987, volume 116, pages 26-37, number 1.[45] Helen Keller,
The Story of My Life , Dover Thrift, 1996.[46] Sara Kiesler,
Fostering Common Ground in Human-Robot Interaction ,Proceedings of the IEEE International Workshop on Robots and HumanInteractive Communication (RO-MAN), 2005, pages 729-734.1047] R. M. Krauss and Susan R. Fussell,
Social psychological models of in-terpersonal communication , Social psychology: Handbook of basic princi-ples, Guilford Press, 1996, editors E. T. Higgins and A. Kruglanski, pages655-701.[48] Robert E. Kraut and Susan R. Fussell and Jane Siegel,
Visual Infor-mation as a Conversational Resource in Collaborative Physical Tasks ,Human-Computer Interaction, 2003, volume 18, pages 13-49.[49] Shuyin Li and Britta Wrede and Gerhard Sagerer,
A computationalmodel of multi-modal grounding , Proceedings of the ACL SIGdial Work-shop on Discourse and Dialog, in conjunction with COLING/ACL 2006,year 2006, pages 153-160, ACL Press.[50] Tim Paek and Eric Horvitz,
Uncertainty, utility, and misunderstanding:A decision-theoretic perspective on grounding in conversational systems ,Psychological models of communication in collaborative systems, Papersfrom the AAAI Fall Symposium, November 5-7, North Falmouth, Mas-sachusett, 1999, pages 85-92.[51] Aaron Powers and Adam D. I. Kramer and Shirlene Lim and Jean Kuoand S. Lee and Sara Kiesler,
Eliciting Information from People with aGendered Humanoid Robot , Proceedings of the IEEE International Work-shop on Robots and Human Interactive Communication (RO-MAN),2005, pages 158-163.[52] Abigail J. Sellen,
Speech Patterns in Video-Mediated Conversations ,Proceedings of the 1992 SIGCHI Conference on Human Factors in Com-puting Systems (CHI 1992), 1992, pages 49-59, ACM.[53] Kerstin Severinson-Eklundh and Helge Huttenrauch and Anders Green,
Social and collaborative aspects of interaction with a service robot. ,Robotics and Autonomous Systems, Special Issue on Socially InteractiveRobots, 2003, volume 42, number 3.[54] Kristen Stubbs and Pamela Hinds and David Wettergreen,
Autonomyand Common Ground in Human-Robot Interaction: A Field Study , IEEEIntelligent Systems, Special Issue on Interacting with Autonomy, 2007,volume 22, pages 42-50, number 2.1155] Kristen Stubbs and Pamela Hinds and David Wettergreen,
Challengesto Grounding in Human-Robot Interaction: Sources of Errors and Mis-communications in Remote Exploration Robotics , Proceedings of the FirstInternational Conference on Human-Robot Interaction, ACM, 2006.[56] Kristen Stubbs and Pamela Hinds and David Wettergreen,
Challengesto Grounding in Human-Robot Collaboration: Errors and Miscommuni-cations in Remote Exploration Robotics , Carnegie Mellon University ,2006.[57] Cristen Torrey and Aaron Powers and Matthew Marge and Susan R.Fussell and Sara Kiesler,
Effects of Adaptive Robot Dialogue on Infor-mation Exchange and Social Relations , Proceedings of the First AnnualConference on Human-Robot Interaction, pages 126-133, ACM, 2006.[58] Veloso, Manuela M.,
Entertainment robotics , Commun. ACM, volume45, number 3, 2002, pages 59-63.[59] Parasuraman, R. and Sheridan, T. B. and Wickens, C. D.,
A model fortypes and levels of human interaction with automation , Trans. Sys. ManCyber. Part A, volume 30, number 3, 2000, pages 286-297.[60] Sinno Jialin Pan and Qiang Yang,