Learn how to implement an obstacle-avoiding path planning for a robot using the Grey Wolf Optimization (GWO) in a static environment. #optimization #algorithm #metaheuristic #robotics #deeplearning #ArtificialIntelligence #MachineLearning #computervision #research #projects #thesis #Python
Get link
Facebook
Twitter
Pinterest
Email
Other Apps
METAHEURISTIC OPTIMIZATION
Get link
Facebook
Twitter
Pinterest
Email
Other Apps
-
Introduction to Metaheuristic Optimization
Metaheuristic represents the family of approximate Optimization Techniques / Algorithms. Using Metaheuristic Techniques we can solve complex real world problems, Engineering design problems, Scientific Problems, Industrial Problems and obtain acceptable solution within time [especially used to solve Science and Engineering Problems].
In Metaheuristic, META means Upper Level Methodology / High Level Procedure and HEURISTIC means Art of discovering new strategies or rules to solve problems / discover or learn something for themselves. Metaheuristic is an advanced technique used to locate, create, or select a heuristic that may provide correct and acceptable results to Optimization Problems.
Optimization is finding better solutions among different possible solutions within acceptable amounts of time. Metaheuristic algorithms are best to solve complex engineering design problems. Popular Metaheuristic Algorithms are:
Genetic Algorithm (GA)
Particle Swarm Optimization Algorithm (PSO)
Ant Colony Optimization Algorithm (ACO)
Artificial Immune System Algorithm (AIS)
Metaheuristic algorithms are used to solve complex problems. Metaheuristic algorithms are not used when efficient exact algorithm can provide solution. For example: It is useless to use metaheuristic algorithms to find out the shortest path in a graph or find minimum spanning tree.
Objective Function f formulate the goal to achieve. Objective function is associated with each solution of the search space [ a real value describe the quality or fitness of the solution]. It is very important to define objective functions properly because Objective functions play major role in designing metaheuristics and guide the search towards the "GOOD" solution of the search space.
HYBRID Metaheuristics are used to solve Multi-Objective Optimization Problems. In Single Objective Optimization unique "GOOD" solution is generated / obtained. In Multi-objective optimization approximated set of Pareto solutions are generated. When dealing with large scale, complex real world optimization problems hybrid metaheuristic provide more efficient and highly flexible solutions.
Hybrid Metaheuristic Optimization Algorithm Example: Mayfly Optimization Algorithm. Mayfly optimization Algorithm is modification of Particle Swarm Optimization (P.S.O). It use the key advantages of Swarm Intelligence Algorithms and Evolutionary Algorithms and form Hybrid Algorithmic Structure. Mayfly optimization Algorithm can be applied on both Continuous and Discrete problems, Sigle Objective and Multi objective optimization problems.
Hybridization drawbacks: New Parameters are introduced to define the hybrid scheme. New parameter setting is nontrivial.
Solution for such problem: Automatic Parameter Setting, adaptive cooperation mechanisms, different approaches such as "COSEARCH" or "Hyperheuristic". Such approaches choose the right heuristic for right operation at the right time during search. Heuristic Space is used to operate Hybrid approaches.
HYPERHEURISITC algorithm is automated methodology is a high level strategy that can control a set of LOW LEVEL HEURISTICS. Hyper-heuristics do not solve problems directly. LOW LEVEL HEURISTICS is selected and applied to the problem. Hyper-heuristic algorithms are used for selection and generate solution to computational problems.
PARTICLE SWARM OPTIMIZATION ALGORITHM NUMERICAL EXAMPLE PSO is a computational method that Optimizes a problem. It is a Population-based stochastic search algorithm. PSO is inspired by the Social Behavior of Birds flocking. n Particle Swarm Optimization the solution of the problem is represented using Particles. [Flocking birds are replaced with particles for algorithm simplicity]. Objective Function is used for the performance evaluation for each particle / agent in the current population. PSO solved problems by having a Population (called Swarms) of Candidate Solutions (Particles). Local and global optimal solutions are used to update particle position in each iteration. Particle Swarm Optimization (PSO) Algorithm step-by-step explanation with Numerical Example and source code implementation. - PART 2 [Example 2] 1.) Initialize Population [Current Iteration (t) = 0] Population Size = 4; 𝑥𝑖 : (i = 1,2,3,4) and (t = 0) 𝑥1 =1.3; 𝑥2=4.3; 𝑥3=0.4; 𝑥4=−1.2 2.) Fitness Function used:
Cuckoo Search Algorithm - Metaheuristic Optimization Algorithm What is Cuckoo Search Algorithm? Cuckoo Search Algorithm is a Meta-Heuristic Algorithm. Cuckoo Search Algorithm is inspired by some Cuckoo species laying their eggs in the nest of other species of birds. In this algorithm, we have 2 bird Species. 1.) Cuckoo birds 2.) Host Birds (Other Species) What if Host Bird discovered cuckoo eggs? Cuckoo eggs can be found by Host Bird. Host bird discovers cuckoos egg with Probability of discovery of alien eggs. If Host Bird Discovered Cuckoo Bird Eggs. The host bird can throw the egg away. Abandon the nest and build a completely new nest. Mathematically, Each egg represent a solution and it is stored in the host bird nest. In this algorithm Artificial Cuckoo Birds are used. Artificial Cuckoo can lay one egg at a time. We will replace New and better solutions with less fit solutions. It means eggs that are more similar to host bird has opportunity to develop in the new generation a
Particle Swarm Optimization (PSO) is a p opulation-based stochastic search algorithm. PSO is inspired by the Social Behavior of Birds flocking. PSO is a computational method that Optimizes a problem. PSO searches for Optima by updating generations. It is popular is an intelligent metaheuristic algorithm. In Particle Swarm Optimization the solution of the problem is represented using Particles. [Flocking birds are replaced with particles for algorithm simplicity]. Objective Function is used for the performance evaluation for each particle / agent in the current population. After a number of iterations agents / particles will find out optimal solution in the search space. Q. What is PSO? A. PSO is a computational method that Optimizes a problem. Q. How PSO will optimize? A. By Improving a Candidate Solution. Q. How PSO Solve Problems? A. PSO solved problems by having a Population (called Swarms) of Candidate Solutions (Particles). Local and global optimal solutions are used to upda
Particle swarm optimization (PSO) What is meant by PSO? PSO is a computational method that Optimizes a problem. It is a Population-based stochastic search algorithm. PSO is inspired by the Social Behavior of Birds flocking. n Particle Swarm Optimization the solution of the problem is represented using Particles. [Flocking birds are replaced with particles for algorithm simplicity]. Objective Function is used for the performance evaluation for each particle / agent in the current population. PSO solved problems by having a Population (called Swarms) of Candidate Solutions (Particles). Local and global optimal solutions are used to update particle position in each iteration. How PSO will optimize? By Improving a Candidate Solution. How PSO Solve Problems? PSO solved problems by having a Population (called Swarms) of Candidate Solutions (Particles). The population of Candidate Solutions (i.e., Particles). What is Search Space in PSO? It is the range in which the algorithm computes the op
Local Binary Pattern Introduction to Local Binary Pattern (LBP) Q. What is Digital Image? A. Digital images are collections of pixels or numbers ( range from 0 to 255). Q. What is Pixel? A. Pixel is the smallest element of any digital image. Pixel can be categorized as Dark Pixel and Bright Pixel. Dark pixels contain low pixel values and bright pixels contain high pixel values. Q. Explain Local Binary Pattern (LBP)? A. Local binary pattern is a popular technique used for image processing. We can use the local binary pattern for face detection and face recognition. Q. What is LBP Operator? A. LBP operator is an image operator. We can transform images into arrays using the LBP operator. Q. How LBP values are computed? A. LBP works in 3x3 (it contain a 9-pixel value ). Local binary pattern looks at nine pixels at a time. Using each 3x3 window in the digital image, we can extract an LBP code. Q. How to Obtain LBP operator value? A. LBP operator values can be obtained by using the simp
There are about 1000 species of Bats. Bat Algorithm is based on the echolocation behavior of Micro Bats with varying pulse rates of emission and loudness. All bats use echolocation to sense distance and background barriers. Microbats are small to medium-sized flying mammals. Micro Bats used a Sonar that is known as Echolocation to detect their prey. Bats fly randomly with the velocity at the position with a fixed frequency and loudness for prey. Q. Whats is Frequency? A. Frequency is the number of waves that pass a fixed point in unit time. Wavelength is the minimum distance between two nearest particles which are in the same phase. Here, Sound waves are used by microbats to detect prey. Q. What is Position? A. A place where something or someone is located. Q. What is Velocity? A. Speed of something in a given direction. Q. What is loudness. A. Loudness refers to how soft or loud sound seems to listeners. Q. What is pulse rate? A. Wave or vibration. In th
Grey Wolf Optimization Algorithm (GWO) Grey Wolf Optimization Grey Wolf Optimization Algorithm is a metaheuristic proposed by Mirjaliali Mohammad and Lewis, 2014. Grey Wolf Optimizer is inspired by the social hierarchy and the hunting technique of Grey Wolves. What is Metaheuristic? Metaheuristic means a High-level problem-independent algorithmic framework (develop optimization algorithms). Metaheuristic algorithms find the best solution out of all possible solutions of optimization. Who are the Grey Wolves? Wolf (Animal): Wolf Lived in a highly organized pack. Also known as Gray wolf or Grey Wolf, is a large canine. Wolf Speed is 50-60 km/h. Their Lifespan is 6-8 years (in the wild). Scientific Name: Canis Lupus. Family: Canidae (Biological family of dog-like carnivorans). Grey Wolves lived in a highly organized pack. The average pack size ranges from 5-12. 4 different ranks of wolves in a pack: Alpha Wolf, Beta Wolf, Delta Wolf, and Omega Wolf. How Grey Wolf Optimization Algorithm
Grey Wolf Optimization Algorithm Numerical Example Grey Wolf Optimization Algorithm Steps 1.) Initialize Grey Wolf Population. 2.) Initialize a, A, and C. 3.) Calculate the fitness of each search agent. 4.) 𝑿_𝜶 = best search agent 5.) 𝑿_𝜷 = second-best search agent 6.) 𝑿_𝜹 = third best search agent. 7.) while (t<Max number of iteration) 8.) For each search agent update the position of the current search agent by the above equations end for 9.) update a, A, and C 10.) Calculate the fitness of all search agents. 11.) update 𝑿_𝜶, 𝑿_𝜷, 𝑿_𝜹 12.) t = t+1 end while 13.) return 𝑿_𝜶 Grey Wolf Optimization Algorithm Numerical Example STEP 1. Initialize the Grey wolf Population [Initial Position for each Search Agent] 𝒙_(𝒊 ) (i = 1,2,3,…n) n = 6 // Number of Search Agents [ -100, 100] // Range Initial Wolf Position 3.2228 4.1553 -3.8197 4.2330 1.3554 -4.1212 STEP 2. Calculate Fitness for Each Search Agent. Objective Function: F6(x) = su
Whale Optimization Algorithm Code Implementation Whale Optimization Algorithm Code Files function obj_fun(test_fun) switch test_fun case 'F1' x = -100:2:100; y=x; case 'F2' x = -10:2:10; y=x; end end function [LB,UB,D,FitFun]=test_fun_info(C) switch C case 'F1' FitFun = @F1; LB = -100; UB = 100; D = 30; case 'F2' FitFun = @F2; LB = -10; UB = 10; D = 30; end % F1 Test Function function r = F1(x) r = sum(x.^2); end % F2 Test Function function r = F2(x) r = sum(abs(x))+prod(abs(x)); end end function Position = initialize(Pop_Size,D,UB,LB) SS_Bounds = size(UB,2); if SS_Bounds == 1 Position = rand(Pop_Size,D).*(UB-LB)+LB; end if SS_Bounds>1 for i = 1:D UB_i = UB(i); LB_i = LB(i); Position(:,i) = rand(Pop_Size,1).*(UB_i-LB_i)+LB_i; end end end function [Best_Val,Best_Pos,Convergence_Curve]=WOA(
Comments
Post a Comment