SlideShare uma empresa Scribd logo
1 de 24
Baixar para ler offline
Amsterdam, The Netherlands
July 06-10, 2013
Ant Colony Optimization
and Swarm Intelligence
Migration Study on a
Pareto-based Island
Model for MOACOs
Antonio M. Mora, P. García-Sánchez,
J.J. Merelo, P.A. Castillo
Depto. de Arquitectura y Tecnología de Computadores
UNIVERSIDAD DE GRANADA
• Ant Colony Optimization
• Multi-objective Optimization
• Island Model
• Pareto-based Island Model
– Model Factors
– Topologies
• MOACS Algorithm
• Experiments and Results
– Attainment Surfaces
– Metrics and Indicators
– Statistics
• Conclusions and Future Work
• Inspired in the behaviour of natural ants when searching for food.
• They cooperate to get the fastest paths between the nest and the source
of food.
• They use a chemical substance named pheromone.
In ACO algorithms
• There are a set of agents called (artificial) ants.
– All of them move in a graph following and depositing (artificial) pheromone.
– They cooperate to find a solution (usually every ant yields a complete
solution).
• There are some formulae applied in the run:
– state transition rule  decides the next step for each ant
– pheromone updating  contribution and evaporation
– evaluation function  assigns the cost to every solution
Ant Colony Optimization
• Optimize several (independent)
objectives at a time.
• There is a set of optimal solutions. Those
which are the best considering all the
objectives than the rest are named non-
dominated solutions.
• The ideal set of non-dominated solutions
is called the Pareto Set (PS).
• Its graphical representation is the Pareto
Front (PF).
Multi-Objective
Optimization
Example of PF for two objective functions.
• If you search in Google (images):
Island Model
• Classical distribution model in EAs:
– Divide or replicate the population in
different subpopulations (islands)
– They evolve independently
– After some generations, some individuals
are chosen in every island
– They migrate to a neighbour island,
following a specific neighbourhood topology
– They replace other individuals in the
receiver islands.
• It Improves the algorithm performance,
not just the running time
Island Model
Island model. Ring neighbourhood topology.
Image by Pablo García-Sánchez
• Just a few works in distributing MOACOs.
• No island model in this type of algorithms, just island model
for ACOs.
• Not take advantage of the multi-colony division both:
– From the algorithmic point of view: very important in Multi-objective
optimization.
– From the computational cost point of view: due to parallelization.
• Previous works by us in this line comparing multi-colony,
sub-colony and island models for different MOACOs.
• Island model with a fixed topology and migration rate.
MOACOs
• The colony is divided into several subcolonies or islands.
• Every island searches in a different area of the space of solutions (PF).
• A l parameter is used for splitting the space in this way.
Model Factors
• Migration policy  the best ants (solutions in fact) are migrated. In this case, the
best regarding the prioritary objective in every island.
• Migrants influence  every migrant contributes to the pheromone matrix of the
receiver island, in order to guide the search to its own area.
• Replacement policy  the migrant is included in the island’s PS, removing those
dominated solutions (by itself) in that set.
• Migration rate  some different values (number of iterations) are tested.
Pareto-based
Island Model
• Migrants move to the closest
neighbour island in the
direction of the prioritary
objective in the colony (sense
of the PF).
• Tries to cover the gaps
between colonies.
Pareto-based
Island Model
Pareto-based
Island Model
• Migrants move to the two
closest neighbour islands in
both directions.
• Aims for a better spread of
solutions in the inter-colony
gaps.
Pareto-based
Island Model
• Every migrant moves from
one island to the rest.
• Aims for a better spread of
solutions along the whole PF.
>>>>>>>>>>>>>>>>>>>
This is a simplified example showing just the
migrations from one island to the rest.
>>>>>>>>>>>>>>>>>>>
• Proposed by Barán et al. in 2003 for solving the VRPTW.
• Proved to be competitive in solving the bicriteria TSP (our test problem)
in previous works.
• It considers one colony, one pheromone matrix and two heuristic
functions (one per objective).
• The state transition rule considers l parameter for weighting the
importance of the heuristic and memoristic (pheromone) information.
• Initially l took a different value per ant in the colony, in order to explore
the whole space of solutions (PF).
• It applies a pheromone reinitialization mechanism, in order to avoid
being stagnated.
MOACS
• 16 islands, everyone in a different processor of a cluster.
• Bicriteria TSP problem: kroAB100, kroAB150 and kroAB200.
• Implemented with MPI (Message Passing Interface).
• 20 runs per experiment.
• q0 and r take values for promoting the exploration more than usual.
• Global PS of all the executions for computing the metrics.
Experiments and Results
• Unidirectional
Experiments and Results
• Bidirectional
Experiments and Results
• Broadcast
Experiments and Results
• Best of every
topology
Experiments and Results
• Hypervolume: it calculates the volume, in the objective space, covered by a set of non-
dominated solutions (PS). A higher value means a better result.
Experiments and Results
• Spread: it measures the extent of spread of a PS. It considers the Euclidean distance between
consecutive solutions on average and extreme distances. A value 0 means an ideal spread.
Experiments and Results
• Epsilon: it is a measure of the smallest distance it would be necessary to translate every
solution in a PS so that it dominates the optimal PF of the problem. Smaller values are better.
Experiments and Results
• Cardinality: number of non-dominated solutions in the obtained PS.
Experiments and Results
• Time: the worst time among all the processors in one execution has been chosen as the
running time of that execution.
Experiments and Results
• A novel island model for MOACOs has been tested with different
topologies and migration rates.
• Bidirectional approach yields a very good balance between quality and
spread in the set of non-dominated solutions (PS).
• It has a flaw concerning the computational time. If it is relevant for the
user, then unidirectional approach would be a better option.
• Regarding the migration rate, high values perform better.
Future Work
• Study some other parameters in the island model.
• Implement and test in other MOACO approaches.
• Compare with state-of-the-art distributed algorithms (EAs, island models)
Conclusions
Questions?
----------------------------------------
Those interested in a partnership with us…
http://geneura.org
THANKYOU!!!!

Mais conteúdo relacionado

Semelhante a Migration study on a Pareto-based island model for MOACOs

seminar HBMO
seminar HBMOseminar HBMO
seminar HBMO
avaninith
 
Multi-Domain Diversity Preservation to Mitigate Particle Stagnation and Enab...
Multi-Domain Diversity Preservation to Mitigate Particle Stagnation and Enab...Multi-Domain Diversity Preservation to Mitigate Particle Stagnation and Enab...
Multi-Domain Diversity Preservation to Mitigate Particle Stagnation and Enab...
Weiyang Tong
 
Feature Selection using Complementary Particle Swarm Optimization for DNA Mic...
Feature Selection using Complementary Particle Swarm Optimization for DNA Mic...Feature Selection using Complementary Particle Swarm Optimization for DNA Mic...
Feature Selection using Complementary Particle Swarm Optimization for DNA Mic...
sky chang
 
Lecture 9 aco
Lecture 9 acoLecture 9 aco
Lecture 9 aco
mcradc
 
CSA 3702 machine learning module 4
CSA 3702 machine learning module 4CSA 3702 machine learning module 4
CSA 3702 machine learning module 4
Nandhini S
 
Particle swarm optimization (PSO) ppt presentation
Particle swarm optimization (PSO) ppt presentationParticle swarm optimization (PSO) ppt presentation
Particle swarm optimization (PSO) ppt presentation
LatestShorts
 
Genetic algorithm_raktim_IITKGP
Genetic algorithm_raktim_IITKGP Genetic algorithm_raktim_IITKGP
Genetic algorithm_raktim_IITKGP
Raktim Halder
 
Ant colony optimization in multiple travelling salesman problem
Ant colony optimization in multiple travelling salesman problemAnt colony optimization in multiple travelling salesman problem
Ant colony optimization in multiple travelling salesman problem
ParishaAgarwal
 

Semelhante a Migration study on a Pareto-based island model for MOACOs (20)

Multi-Verse Optimizer
Multi-Verse OptimizerMulti-Verse Optimizer
Multi-Verse Optimizer
 
seminar HBMO
seminar HBMOseminar HBMO
seminar HBMO
 
Multi-Domain Diversity Preservation to Mitigate Particle Stagnation and Enab...
Multi-Domain Diversity Preservation to Mitigate Particle Stagnation and Enab...Multi-Domain Diversity Preservation to Mitigate Particle Stagnation and Enab...
Multi-Domain Diversity Preservation to Mitigate Particle Stagnation and Enab...
 
Ant Colony Optimization (ACO)
Ant Colony Optimization (ACO)Ant Colony Optimization (ACO)
Ant Colony Optimization (ACO)
 
Ant Colony Optimization algorithms in ADSA
Ant Colony Optimization algorithms in ADSAAnt Colony Optimization algorithms in ADSA
Ant Colony Optimization algorithms in ADSA
 
SPLT Transformer.pptx
SPLT Transformer.pptxSPLT Transformer.pptx
SPLT Transformer.pptx
 
Updating Ecological Niche Modeling Methodologies
Updating Ecological Niche Modeling MethodologiesUpdating Ecological Niche Modeling Methodologies
Updating Ecological Niche Modeling Methodologies
 
D1T3 enm workflows updated
D1T3 enm workflows updatedD1T3 enm workflows updated
D1T3 enm workflows updated
 
Feature Selection using Complementary Particle Swarm Optimization for DNA Mic...
Feature Selection using Complementary Particle Swarm Optimization for DNA Mic...Feature Selection using Complementary Particle Swarm Optimization for DNA Mic...
Feature Selection using Complementary Particle Swarm Optimization for DNA Mic...
 
Lecture 9 aco
Lecture 9 acoLecture 9 aco
Lecture 9 aco
 
A survey on ant colony clustering papers
A survey on ant colony clustering papersA survey on ant colony clustering papers
A survey on ant colony clustering papers
 
An Updated Survey on Niching Methods and Their Applications
An Updated Survey on Niching Methods and Their ApplicationsAn Updated Survey on Niching Methods and Their Applications
An Updated Survey on Niching Methods and Their Applications
 
CSA 3702 machine learning module 4
CSA 3702 machine learning module 4CSA 3702 machine learning module 4
CSA 3702 machine learning module 4
 
Particle swarm optimization (PSO) ppt presentation
Particle swarm optimization (PSO) ppt presentationParticle swarm optimization (PSO) ppt presentation
Particle swarm optimization (PSO) ppt presentation
 
27.docking protein-protein and protein-ligand
27.docking protein-protein and protein-ligand27.docking protein-protein and protein-ligand
27.docking protein-protein and protein-ligand
 
A multivariate approach for process variograms
A multivariate approach for process variogramsA multivariate approach for process variograms
A multivariate approach for process variograms
 
Genetic algorithm raktim
Genetic algorithm raktimGenetic algorithm raktim
Genetic algorithm raktim
 
Genetic algorithm_raktim_IITKGP
Genetic algorithm_raktim_IITKGP Genetic algorithm_raktim_IITKGP
Genetic algorithm_raktim_IITKGP
 
Ant colony optimization in multiple travelling salesman problem
Ant colony optimization in multiple travelling salesman problemAnt colony optimization in multiple travelling salesman problem
Ant colony optimization in multiple travelling salesman problem
 
AI Robotics
AI RoboticsAI Robotics
AI Robotics
 

Mais de Antonio Mora

Optimización Adaptativa basada en Colonias de Hormigas para la Composición de...
Optimización Adaptativa basada en Colonias de Hormigas para la Composición de...Optimización Adaptativa basada en Colonias de Hormigas para la Composición de...
Optimización Adaptativa basada en Colonias de Hormigas para la Composición de...
Antonio Mora
 
GRETIVE: Un Bot Evolutivo para HearthStone basado en Perfiles
GRETIVE: Un Bot Evolutivo para HearthStone basado en PerfilesGRETIVE: Un Bot Evolutivo para HearthStone basado en Perfiles
GRETIVE: Un Bot Evolutivo para HearthStone basado en Perfiles
Antonio Mora
 
Beating uncertainty in racing bot evolution through enhanced exploration and ...
Beating uncertainty in racing bot evolution through enhanced exploration and ...Beating uncertainty in racing bot evolution through enhanced exploration and ...
Beating uncertainty in racing bot evolution through enhanced exploration and ...
Antonio Mora
 
The Evolutionary Race: Improving the Process of Evaluating Car Controllers in...
The Evolutionary Race: Improving the Process of Evaluating Car Controllers in...The Evolutionary Race: Improving the Process of Evaluating Car Controllers in...
The Evolutionary Race: Improving the Process of Evaluating Car Controllers in...
Antonio Mora
 
Sólo puede quedar uno: Evolución de Bots para RTS basada en supervivencia
Sólo puede quedar uno: Evolución de Bots para RTS basada en supervivenciaSólo puede quedar uno: Evolución de Bots para RTS basada en supervivencia
Sólo puede quedar uno: Evolución de Bots para RTS basada en supervivencia
Antonio Mora
 
Living-UGR: Una aventura gráfica geolocalizada para difundir el patrimonio de...
Living-UGR: Una aventura gráfica geolocalizada para difundir el patrimonio de...Living-UGR: Una aventura gráfica geolocalizada para difundir el patrimonio de...
Living-UGR: Una aventura gráfica geolocalizada para difundir el patrimonio de...
Antonio Mora
 

Mais de Antonio Mora (20)

Study on Genetic Algorithm Approaches to Improve an Autonomous Agent for a Fi...
Study on Genetic Algorithm Approaches to Improve an Autonomous Agent for a Fi...Study on Genetic Algorithm Approaches to Improve an Autonomous Agent for a Fi...
Study on Genetic Algorithm Approaches to Improve an Autonomous Agent for a Fi...
 
Optimización Adaptativa basada en Colonias de Hormigas para la Composición de...
Optimización Adaptativa basada en Colonias de Hormigas para la Composición de...Optimización Adaptativa basada en Colonias de Hormigas para la Composición de...
Optimización Adaptativa basada en Colonias de Hormigas para la Composición de...
 
Adaptive Ant Colony Optimization for Service Function Chaining in a Dynamic 5...
Adaptive Ant Colony Optimization for Service Function Chaining in a Dynamic 5...Adaptive Ant Colony Optimization for Service Function Chaining in a Dynamic 5...
Adaptive Ant Colony Optimization for Service Function Chaining in a Dynamic 5...
 
Research in Videogames. (Much) further than just AI
Research in Videogames. (Much) further than just AIResearch in Videogames. (Much) further than just AI
Research in Videogames. (Much) further than just AI
 
GRETIVE: Un Bot Evolutivo para HearthStone basado en Perfiles
GRETIVE: Un Bot Evolutivo para HearthStone basado en PerfilesGRETIVE: Un Bot Evolutivo para HearthStone basado en Perfiles
GRETIVE: Un Bot Evolutivo para HearthStone basado en Perfiles
 
Improving the Performance of MCTS-Based μRTS Agents Through Move Pruning
Improving the Performance of MCTS-Based μRTS Agents Through Move PruningImproving the Performance of MCTS-Based μRTS Agents Through Move Pruning
Improving the Performance of MCTS-Based μRTS Agents Through Move Pruning
 
Testing hybrid computational intelligence algorithms for general game playing...
Testing hybrid computational intelligence algorithms for general game playing...Testing hybrid computational intelligence algorithms for general game playing...
Testing hybrid computational intelligence algorithms for general game playing...
 
Checking the difficulty of evolutionary-generated maps in a N-Body inspired m...
Checking the difficulty of evolutionary-generated maps in a N-Body inspired m...Checking the difficulty of evolutionary-generated maps in a N-Body inspired m...
Checking the difficulty of evolutionary-generated maps in a N-Body inspired m...
 
Applying Ant Colony Optimization for Service Function Chaining in a 5G Networ...
Applying Ant Colony Optimization for Service Function Chaining in a 5G Networ...Applying Ant Colony Optimization for Service Function Chaining in a 5G Networ...
Applying Ant Colony Optimization for Service Function Chaining in a 5G Networ...
 
Investigación en videojuegos. (mucho) Mas allá de la IA
Investigación en videojuegos. (mucho) Mas allá de la IAInvestigación en videojuegos. (mucho) Mas allá de la IA
Investigación en videojuegos. (mucho) Mas allá de la IA
 
Beating uncertainty in racing bot evolution through enhanced exploration and ...
Beating uncertainty in racing bot evolution through enhanced exploration and ...Beating uncertainty in racing bot evolution through enhanced exploration and ...
Beating uncertainty in racing bot evolution through enhanced exploration and ...
 
Inteligencia Computacional en Videojuegos (Meetup GranadAI 2019)
Inteligencia Computacional en Videojuegos (Meetup GranadAI 2019)Inteligencia Computacional en Videojuegos (Meetup GranadAI 2019)
Inteligencia Computacional en Videojuegos (Meetup GranadAI 2019)
 
Free Form Evolution for Angry Birds Level Generation
Free Form Evolution for Angry Birds Level GenerationFree Form Evolution for Angry Birds Level Generation
Free Form Evolution for Angry Birds Level Generation
 
Ciencia y Videojuegos (ULP 2019)
Ciencia y Videojuegos (ULP 2019)Ciencia y Videojuegos (ULP 2019)
Ciencia y Videojuegos (ULP 2019)
 
The Evolutionary Race: Improving the Process of Evaluating Car Controllers in...
The Evolutionary Race: Improving the Process of Evaluating Car Controllers in...The Evolutionary Race: Improving the Process of Evaluating Car Controllers in...
The Evolutionary Race: Improving the Process of Evaluating Car Controllers in...
 
Predicción de Quiebra Financiera de Empresas Mediante Equilibrado de Datos y ...
Predicción de Quiebra Financiera de Empresas Mediante Equilibrado de Datos y ...Predicción de Quiebra Financiera de Empresas Mediante Equilibrado de Datos y ...
Predicción de Quiebra Financiera de Empresas Mediante Equilibrado de Datos y ...
 
Driving in TORCS using modular fuzzy controllers - Poster - EvoGAMES 2017
Driving in TORCS using modular fuzzy controllers - Poster - EvoGAMES 2017Driving in TORCS using modular fuzzy controllers - Poster - EvoGAMES 2017
Driving in TORCS using modular fuzzy controllers - Poster - EvoGAMES 2017
 
Sólo puede quedar uno: Evolución de Bots para RTS basada en supervivencia
Sólo puede quedar uno: Evolución de Bots para RTS basada en supervivenciaSólo puede quedar uno: Evolución de Bots para RTS basada en supervivencia
Sólo puede quedar uno: Evolución de Bots para RTS basada en supervivencia
 
Living-UGR: Una aventura gráfica geolocalizada para difundir el patrimonio de...
Living-UGR: Una aventura gráfica geolocalizada para difundir el patrimonio de...Living-UGR: Una aventura gráfica geolocalizada para difundir el patrimonio de...
Living-UGR: Una aventura gráfica geolocalizada para difundir el patrimonio de...
 
Gamification in Teaching - How to motivate students through games
Gamification in Teaching - How to motivate students through gamesGamification in Teaching - How to motivate students through games
Gamification in Teaching - How to motivate students through games
 

Último

Histor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slideHistor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slide
vu2urc
 
Artificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and MythsArtificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and Myths
Joaquim Jorge
 

Último (20)

From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
 
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdfUnderstanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
 
How to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected WorkerHow to Troubleshoot Apps for the Modern Connected Worker
How to Troubleshoot Apps for the Modern Connected Worker
 
Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...Driving Behavioral Change for Information Management through Data-Driven Gree...
Driving Behavioral Change for Information Management through Data-Driven Gree...
 
Developing An App To Navigate The Roads of Brazil
Developing An App To Navigate The Roads of BrazilDeveloping An App To Navigate The Roads of Brazil
Developing An App To Navigate The Roads of Brazil
 
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
 
Histor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slideHistor y of HAM Radio presentation slide
Histor y of HAM Radio presentation slide
 
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, AdobeApidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
 
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
Mastering MySQL Database Architecture: Deep Dive into MySQL Shell and MySQL R...
 
Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024Axa Assurance Maroc - Insurer Innovation Award 2024
Axa Assurance Maroc - Insurer Innovation Award 2024
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
Handwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed textsHandwritten Text Recognition for manuscripts and early printed texts
Handwritten Text Recognition for manuscripts and early printed texts
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
 
What Are The Drone Anti-jamming Systems Technology?
What Are The Drone Anti-jamming Systems Technology?What Are The Drone Anti-jamming Systems Technology?
What Are The Drone Anti-jamming Systems Technology?
 
presentation ICT roal in 21st century education
presentation ICT roal in 21st century educationpresentation ICT roal in 21st century education
presentation ICT roal in 21st century education
 
[2024]Digital Global Overview Report 2024 Meltwater.pdf
[2024]Digital Global Overview Report 2024 Meltwater.pdf[2024]Digital Global Overview Report 2024 Meltwater.pdf
[2024]Digital Global Overview Report 2024 Meltwater.pdf
 
Scaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organizationScaling API-first – The story of a global engineering organization
Scaling API-first – The story of a global engineering organization
 
A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?
 
Artificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and MythsArtificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and Myths
 
Tech Trends Report 2024 Future Today Institute.pdf
Tech Trends Report 2024 Future Today Institute.pdfTech Trends Report 2024 Future Today Institute.pdf
Tech Trends Report 2024 Future Today Institute.pdf
 

Migration study on a Pareto-based island model for MOACOs

  • 1. Amsterdam, The Netherlands July 06-10, 2013 Ant Colony Optimization and Swarm Intelligence Migration Study on a Pareto-based Island Model for MOACOs Antonio M. Mora, P. García-Sánchez, J.J. Merelo, P.A. Castillo Depto. de Arquitectura y Tecnología de Computadores UNIVERSIDAD DE GRANADA
  • 2. • Ant Colony Optimization • Multi-objective Optimization • Island Model • Pareto-based Island Model – Model Factors – Topologies • MOACS Algorithm • Experiments and Results – Attainment Surfaces – Metrics and Indicators – Statistics • Conclusions and Future Work
  • 3. • Inspired in the behaviour of natural ants when searching for food. • They cooperate to get the fastest paths between the nest and the source of food. • They use a chemical substance named pheromone. In ACO algorithms • There are a set of agents called (artificial) ants. – All of them move in a graph following and depositing (artificial) pheromone. – They cooperate to find a solution (usually every ant yields a complete solution). • There are some formulae applied in the run: – state transition rule  decides the next step for each ant – pheromone updating  contribution and evaporation – evaluation function  assigns the cost to every solution Ant Colony Optimization
  • 4. • Optimize several (independent) objectives at a time. • There is a set of optimal solutions. Those which are the best considering all the objectives than the rest are named non- dominated solutions. • The ideal set of non-dominated solutions is called the Pareto Set (PS). • Its graphical representation is the Pareto Front (PF). Multi-Objective Optimization Example of PF for two objective functions.
  • 5. • If you search in Google (images): Island Model
  • 6. • Classical distribution model in EAs: – Divide or replicate the population in different subpopulations (islands) – They evolve independently – After some generations, some individuals are chosen in every island – They migrate to a neighbour island, following a specific neighbourhood topology – They replace other individuals in the receiver islands. • It Improves the algorithm performance, not just the running time Island Model Island model. Ring neighbourhood topology. Image by Pablo García-Sánchez
  • 7. • Just a few works in distributing MOACOs. • No island model in this type of algorithms, just island model for ACOs. • Not take advantage of the multi-colony division both: – From the algorithmic point of view: very important in Multi-objective optimization. – From the computational cost point of view: due to parallelization. • Previous works by us in this line comparing multi-colony, sub-colony and island models for different MOACOs. • Island model with a fixed topology and migration rate. MOACOs
  • 8. • The colony is divided into several subcolonies or islands. • Every island searches in a different area of the space of solutions (PF). • A l parameter is used for splitting the space in this way. Model Factors • Migration policy  the best ants (solutions in fact) are migrated. In this case, the best regarding the prioritary objective in every island. • Migrants influence  every migrant contributes to the pheromone matrix of the receiver island, in order to guide the search to its own area. • Replacement policy  the migrant is included in the island’s PS, removing those dominated solutions (by itself) in that set. • Migration rate  some different values (number of iterations) are tested. Pareto-based Island Model
  • 9. • Migrants move to the closest neighbour island in the direction of the prioritary objective in the colony (sense of the PF). • Tries to cover the gaps between colonies. Pareto-based Island Model
  • 10. Pareto-based Island Model • Migrants move to the two closest neighbour islands in both directions. • Aims for a better spread of solutions in the inter-colony gaps.
  • 11. Pareto-based Island Model • Every migrant moves from one island to the rest. • Aims for a better spread of solutions along the whole PF. >>>>>>>>>>>>>>>>>>> This is a simplified example showing just the migrations from one island to the rest. >>>>>>>>>>>>>>>>>>>
  • 12. • Proposed by Barán et al. in 2003 for solving the VRPTW. • Proved to be competitive in solving the bicriteria TSP (our test problem) in previous works. • It considers one colony, one pheromone matrix and two heuristic functions (one per objective). • The state transition rule considers l parameter for weighting the importance of the heuristic and memoristic (pheromone) information. • Initially l took a different value per ant in the colony, in order to explore the whole space of solutions (PF). • It applies a pheromone reinitialization mechanism, in order to avoid being stagnated. MOACS
  • 13. • 16 islands, everyone in a different processor of a cluster. • Bicriteria TSP problem: kroAB100, kroAB150 and kroAB200. • Implemented with MPI (Message Passing Interface). • 20 runs per experiment. • q0 and r take values for promoting the exploration more than usual. • Global PS of all the executions for computing the metrics. Experiments and Results
  • 17. • Best of every topology Experiments and Results
  • 18. • Hypervolume: it calculates the volume, in the objective space, covered by a set of non- dominated solutions (PS). A higher value means a better result. Experiments and Results
  • 19. • Spread: it measures the extent of spread of a PS. It considers the Euclidean distance between consecutive solutions on average and extreme distances. A value 0 means an ideal spread. Experiments and Results
  • 20. • Epsilon: it is a measure of the smallest distance it would be necessary to translate every solution in a PS so that it dominates the optimal PF of the problem. Smaller values are better. Experiments and Results
  • 21. • Cardinality: number of non-dominated solutions in the obtained PS. Experiments and Results
  • 22. • Time: the worst time among all the processors in one execution has been chosen as the running time of that execution. Experiments and Results
  • 23. • A novel island model for MOACOs has been tested with different topologies and migration rates. • Bidirectional approach yields a very good balance between quality and spread in the set of non-dominated solutions (PS). • It has a flaw concerning the computational time. If it is relevant for the user, then unidirectional approach would be a better option. • Regarding the migration rate, high values perform better. Future Work • Study some other parameters in the island model. • Implement and test in other MOACO approaches. • Compare with state-of-the-art distributed algorithms (EAs, island models) Conclusions
  • 24. Questions? ---------------------------------------- Those interested in a partnership with us… http://geneura.org THANKYOU!!!!