Improved Search for Local Optima in Particle Swarm Optimization May 6, 2015 Huidae Cho Water Resources Engineer, Dewberry Consultants Part-Time Assistant.

Slides:



Advertisements
Similar presentations
Bayesian Belief Propagation
Advertisements

Local optimization technique G.Anuradha. Introduction The evaluation function defines a quality measure score landscape/response surface/fitness landscape.
1 An Adaptive GA for Multi Objective Flexible Manufacturing Systems A. Younes, H. Ghenniwa, S. Areibi uoguelph.ca.
Particle Swarm Optimization
A Hybrid IWO/PSO Algorithm for Fast and Global Optimization Hossein Hajimirsadeghi.
All Hands Meeting, 2006 Title: Grid Workflow Scheduling in WOSE (Workflow Optimisation Services for e- Science Applications) Authors: Yash Patel, Andrew.
Particle Swarm Optimization (PSO)
Gizem ALAGÖZ. Simulation optimization has received considerable attention from both simulation researchers and practitioners. Both continuous and discrete.
Spring, 2013C.-S. Shieh, EC, KUAS, Taiwan1 Heuristic Optimization Methods Pareto Multiobjective Optimization Patrick N. Ngatchou, Anahita Zarei, Warren.
Estimation of Distribution Algorithms Ata Kaban School of Computer Science The University of Birmingham.
EMBIO – Cambridge Particle Swarm Optimization applied to Automated Docking Automated docking of a ligand to a macromolecule Particle Swarm Optimization.
MAE 552 – Heuristic Optimization Lecture 6 February 6, 2002.
The information contained in this document pertains to software products and services that are subject to the controls of the Export Administration Regulations.
Reporter : Mac Date : Multi-Start Method Rafael Marti.
1 A hybrid particle swarm optimization algorithm for optimal task assignment in distributed system Peng-Yeng Yin and Pei-Pei Wang Department of Information.
Evolutionary Computational Intelligence Lecture 9: Noisy Fitness Ferrante Neri University of Jyväskylä.
Parameterizing Random Test Data According to Equivalence Classes Chris Murphy, Gail Kaiser, Marta Arias Columbia University.
Ant Colony Optimization to Resource Allocation Problems Peng-Yeng Yin and Ching-Yu Wang Department of Information Management National Chi Nan University.
Advanced Topics in Optimization
K-means Clustering. What is clustering? Why would we want to cluster? How would you determine clusters? How can you do this efficiently?
Nonlinear Stochastic Programming by the Monte-Carlo method Lecture 4 Leonidas Sakalauskas Institute of Mathematics and Informatics Vilnius, Lithuania EURO.
Stochastic Approximation and Simulated Annealing Lecture 8 Leonidas Sakalauskas Institute of Mathematics and Informatics Vilnius, Lithuania EURO Working.
Multimodal Optimization (Niching) A/Prof. Xiaodong Li School of Computer Science and IT, RMIT University Melbourne, Australia
FDA- A scalable evolutionary algorithm for the optimization of ADFs By Hossein Momeni.
1 Institute of Engineering Mechanics Leopold-Franzens University Innsbruck, Austria, EU H.J. Pradlwarter and G.I. Schuëller Confidence.
Optimum Design of Steel Space Frames by Hybrid Teaching-Learning Based Optimization and Harmony Search Algorithms & Dr.Alper AKIN Dr. IbrahIm AYDOGDU Dear.
Hierarchical Distributed Genetic Algorithm for Image Segmentation Hanchuan Peng, Fuhui Long*, Zheru Chi, and Wanshi Siu {fhlong, phc,
Swarm Intelligence 虞台文.
Stochastic Linear Programming by Series of Monte-Carlo Estimators Leonidas SAKALAUSKAS Institute of Mathematics&Informatics Vilnius, Lithuania
Particle Swarm Optimization (PSO) Algorithm and Its Application in Engineering Design Optimization School of Information Technology Indian Institute of.
May 6, 2015 Huidae Cho Water Resources Engineer, Dewberry Consultants
(Particle Swarm Optimisation)
The Particle Swarm Optimization Algorithm Nebojša Trpković 10 th Dec 2010.
Topics in Artificial Intelligence By Danny Kovach.
Simulated Annealing.
Mathematical Models & Optimization?
FORS 8450 Advanced Forest Planning Lecture 11 Tabu Search.
Solving of Graph Coloring Problem with Particle Swarm Optimization Amin Fazel Sharif University of Technology Caro Lucas February 2005 Computer Engineering.
1 Effect of Spatial Locality on An Evolutionary Algorithm for Multimodal Optimization EvoNum 2010 Ka-Chun Wong, Kwong-Sak Leung, and Man-Hon Wong Department.
Biologically inspired algorithms BY: Andy Garrett YE Ziyu.
Particle Swarm Optimization for Image Analysis Stefano Cagnoni Dipartimento di Ingegneria dell’Informazione Università di Parma.
Particle Swarm Optimization by Dr. Shubhajit Roy Chowdhury Centre for VLSI and Embedded Systems Technology, IIIT Hyderabad.
Particle Swarm Optimization † Spencer Vogel † This presentation contains cheesy graphics and animations and they will be awesome.
De novo discovery of mutated driver pathways in cancer Discussion leader: Matthew Bernstein Scribe: Kun-Chieh Wang Computational Network Biology BMI 826/Computer.
Data Mining and Decision Support
September 28, 2000 Improved Simultaneous Data Reconciliation, Bias Detection and Identification Using Mixed Integer Optimization Methods Presented by:
An Introduction to Simulated Annealing Kevin Cannons November 24, 2005.
Faculty of Information Engineering, Shenzhen University Liao Huilian SZU TI-DSPs LAB Aug 27, 2007 Optimizer based on particle swarm optimization and LBG.
Generalized Point Based Value Iteration for Interactive POMDPs Prashant Doshi Dept. of Computer Science and AI Institute University of Georgia
An Improved Quantum-behaved Particle Swarm Optimization Algorithm Based on Culture V i   v i 1, v i 2,.. v iD  Gao X. Z 2, Wu Ying 1, Huang Xianlin.
Metaheuristics for the New Millennium Bruce L. Golden RH Smith School of Business University of Maryland by Presented at the University of Iowa, March.
Non-parametric Methods for Clustering Continuous and Categorical Data Steven X. Wang Dept. of Math. and Stat. York University May 13, 2010.
Let-It-Rain: A Web-based Stochastic Rainfall Generator Huidae Cho 1 Dekay Kim 2, Christian Onof 3, Minha Choi 4 April 20, Dewberry, Atlanta, GA.
A Hybrid Optimization Approach for Automated Parameter Estimation Problems Carlos A. Quintero 1 Miguel Argáez 1, Hector Klie 2, Leticia Velázquez 1 and.
Applications of Tabu Search OPIM 950 Gary Chen 9/29/03.
On the Computation of All Global Minimizers Through Particle Swarm Optimization IEEE Transactions On Evolutionary Computation, Vol. 8, No.3, June 2004.
Particle Swarm Optimization (PSO) Algorithm. Swarming – The Definition aggregation of similar animals, generally cruising in the same directionaggregation.
Swarm Intelligence. Content Overview Swarm Particle Optimization (PSO) – Example Ant Colony Optimization (ACO)
Particle Swarm Optimization (2)
Heuristic Optimization Methods
Adnan Quadri & Dr. Naima Kaabouch Optimization Efficiency
Particle Swarm Optimization
Particle Swarm Optimization
PSO -Introduction Proposed by James Kennedy & Russell Eberhart in 1995
Meta-heuristics Introduction - Fabien Tricoire
آموزش شبکه عصبی با استفاده از روش بهینه سازی PSO
Weihua Gao Ganapathi Kamath Kalyan Veeramachaneni Lisa Osadciw
OPTIMIZATION OF PLANAR TRUSS STRUCTURE USING FIREFLY ALGORITHM
Multi-band impedance matching using an evolutionary algorithm
现代智能优化算法-粒子群算法 华北电力大学输配电系统研究所 刘自发 2008年3月 1/18/2019
Presentation transcript:

Improved Search for Local Optima in Particle Swarm Optimization May 6, 2015 Huidae Cho Water Resources Engineer, Dewberry Consultants Part-Time Assistant Professor, Kennesaw State University

2 Overview Why Find Local Optima? Isolated-Speciation-based Particle Swarm Optimization (ISPSO) Challenges in Multi-Modal Optimization Stochastic Rainfall Generator Other Applications Conclusions

3 Why Find Local Optima? Traditional model optimization tries to find “ the ” global optimum only. Is the/a global optimum always what we want? We want “ realistic ” solutions. Need a new technique to find many working solutions.

4 Why Find Local Optima? (Cont.) Flood risk model Want to minimize the risk. Finds only the global optimum. If factors A and B are costly? Factor A Factor B Risk

5 Isolated-Speciation-based Particle Swarm Optimization (ISPSO) New optimization method based on Species-based PSO (SPSO)! Implemented in the R language. Runs on multi-platforms: MS-Windows, UNIX Finds local optima as well as the global optimum.

6 ISPSO (Cont.) Particle Swarm Optimization (PSO) PSO is a metaheuristic based on the movement of possible solutions referred to as particles. A swarm consists of multiple particles sharing information with each other. Particles fly through the search space towards optimal solutions.

7 ISPSO (Cont.) Randomness vs. Uniformity

8 ISPSO (Cont.) Particle ’ s Movement ~ v i ( t ) ~ x i ( t )  ~ v i ( t ) x 2 ~ p i x 1 ~ p g ~ x i ( t + 1 ) ~ v i ( t + 1 )  à 1 ~ r 1 ( t )( ~ p i ¡ ~ x i ( t ))  à 2 ~ r 2 ( t )( ~ p g ¡ ~ x i ( t ))  = 2 j 2 ¡ à ¡ p à 2 ¡ 4 à j ; à = à 1 + à 2 > 4 ~ v i ( t + 1 ) =  [ ~ v i ( t ) + à 1 ~ r 1 ( t )( ~ p i ¡ ~ x i ( t )) + à 2 ~ r 2 ( t )( ~ p g ¡ ~ x i ( t ))] ~ x i ( t + 1 ) = ~ x i ( t ) + ~ v i ( t + 1 )

9 ISPSO (Cont.) Isolated Speciation

10 ISPSO (Cont.) Fitness Assimilation

11 ISPSO (Cont.) Preemptive Competitive Nesting

12 ISPSO (Cont.) Flowchart Initial population from Sobol’ sequences Isolated speciation Update velocities Check for nesting criteria Preemptive nesting Update positions Fitness assimilation Stopping criteria? No Yes Start End

13 ISPSO (Cont.) ISPSO vs. SPSO

14 Challenges in Multi-Modal Optimization How to Detect the Findings of Local Optima? SPSO: error between the fitness of the global optimum and the particle ’ s fitness  In real-world problems, in most cases, the fitness of the global optimum unknown. NichePSO: the standard deviation of fitness values over a number of iteration  Cannot guarantee spatial convergence. ISPSO: SD of fitness and the normalized geometric mean of the particle ’ s half-life experience.  Guarantees spatial convergence as well as fitness convergence.

15 Challenges in Multi-Modal Optimization (Cont.) When Failed to Detect Local Optima Correctly? nAnalytically counted by Cho et al. (2008), Numerically confirmed with ISPSO NichePSO by Brits et al. (2007) , Number of Global and Local Optima in [-28,28] n

16 Challenges in Multi-Modal Optimization (Cont.) When to Stop the Algorithm? Don ’ t want to wait until the maximum number of iterations. As the number of iterations between successive discoveries increases  It becomes more difficult to find more optima.  Possibility to find another optimum decreases.  Certain threshold.

17 Stochastic Rainfall Generator Modified Bartlett-Lewis Rectangular Pulse Model (MBLRP) Stochastically generates synthetic rainfall time series. Replicates statistics of observed rainfall at a given rain gage. Highly multi-modal: Need to find as many feasible solutions as possible.

18 Stochastic Rainfall Generator (Cont.) ISPSO vs. NichePSO: 2D Projection of Solutions

19 Stochastic Rainfall Generator (Cont.) ISPSO vs. NichePSO: Histograms of the Normalized Distance between Solutions and True Optima

20 ISPSO vs. AutoCal built in SWAT Other Applications SWAT Model Calibration Clear Multi-Modality!

21 Other Applications (Cont.) Uncertainty Analysis within the GLUE Framework Multi-modal optimization suitable for equifinality Relative sampling density In a SWAT case study, 4,000 model runs vs. 46,000!

22 References Brits, R., Engelbrecht, A.P., van den Bergh, F., Locating Multiple Optima Using Particle Swarm Optimization. Applied Mathematics and Computation 189 (2), Cho, H., Kim, D., Olivera, F., Guikema, S. D., Enhanced Speciation in Particle Swarm Optimization for Multi-Modal Problems. European Journal of Operational Research 213 (1), Cho, H., Olivera, F., Guikema, S. D., A Derivation of the Number of Minima of the Griewank Function. Applied Mathematics and Computation 204 (2),

23 Conclusions Mathematically best solutions are not always practical and feasible. ISPSO improved SPSO for finding local optima. More reliable criteria for finding solutions and stopping optimization were introduced. ISPSO outperformed SPSO, its predecessor, and NichePSO, another PSO-based multi-modal optimizer. Application to model calibration and GLUE.