运动科学综述阅读(2): Swarm Intelligence,Physically Inspired algorithms

上回涉及introduction，主要包含evolutionary algorithms。 本回介绍swarm intelligence和physically inspired algorithms(ps. 主要就是模拟退火）
SI
PSO
inspired by the optimized behavior of bird flocks and fish schools
Particle: a posslible solution of a given problem, flies in the search space according to its previous best position(PBest), so far the best position of the flock(gbest) and its velocity at the moment.
basically a particle knows four facts about its envrironment:
1.its position and velocity at a given time
2.the fitness of its current position
3.the best position it has ever achieved, and the value associated with it(PBest)
4.its neighbors, their best positions and the values(gbest)
How to calculate velocity of a given time?
1.inertia component: keeping the particle moving in the same direction it was originally heading
2.cognitive component: acts as the particle’s memory, causing it to tend to return to the regions of the search space in which it has experienced higher individual fitness
3.social component: causes the particle to move to the best region the swarm has found so far
velocity determines the particle’s next position
Coello et al., 2004,
Parsopoulos and Vrahatis, 2002

FA
a population of fireflies representing a feasible set of solutions to the problem
Each firefly has a fitness value and according to the fitness, the fireflies (solutions) in the population move towards better fireflies. The movement function in the canonical algorithm consists of the current firefly and the better firefly, the attraction between the two fireflies and a stochastic component.
Senthilnath et al., 2011
Yang and He, 2013

ACO
inspired by the foraging behavior of ants, pheromone communication. … this mechanism is called stigmergy.

共识主动性（英语：Stigmergy）是社会网络中生物个体自治的信息协调机制。在没有中枢控制和接触交流的条件下，群体通过同频共振，达到信息对称，个体独立行动，互相修正，自我更新，逐步完善群体的生态环境。

Start with positioning ants(defining starting points)
application: path optimization, scheduling tasks
Gao et al., 2013
Gomez et al., 2004

CSA
inspired by obligate brood parasitism of some cuckoo species by laying their eggs in the nest of host birds
no distinctino between a nest, an egg or a cuckoo; a solution represents them all
each egg in a nest represents a solution, and a cuckoo egg represents a new solution
The aim is to employ the new and potentially better solutions (cuckoos) to replace the not-so-good solutions in the nests.
Ouaarab et al., 2014
Yang and Deb, 2010

ASFO
artificial fish swarm optimization
qian, 2002
incorporates social behaviour of natural swarms and the concepts in evolutionary computing

exchange data with their adjacent members
●prey: randomly swim in search of food
●follow: when a fish finds a location with abundant food and other fish quickly follow
●swarm: the tendency for fish to gather in groups while swimming

faster convergence ability and the availability of the less parameters which increase the usability
Cheng et al., 2009
Neshat et al., 2014

EPO
emperor penguin optimizer, Dhiman and Kumar, 2018
inspired by the optimized huddling behaviour of emperor penguins
The penguins will move finding better positions based on the huddle boundary, temperature profile and the distance between the penguins in the population.
engineering optimization tasks
Baliarsingh et al., 2019
Dhiman et al., 2021

SSO
spider-squirrel optimization
hybrid from two SI algorithms: spider-monkey optimization and squirrel search algorithms
SMO:
Bansal et al., 2014
follow fission- fusion social systems as basis
mimic the food finding strategy of monkeys by dividing into sub groups
SSA:
Jain et al., 2019
SSA is motivated by the grouping behaviour of southern flying squirrels and the way they glide
SSO diminish the drawback of SMO such as minimal convergence by combining the SSA which is able to effectively find the global optimum solution with faster convergence

TLBO
teaching-learning-based optimization
Rao et al., 2011
想要解决调参的问题, develop a parameter free algorithm

best learner as teacher: as a teacher is considered to be a well learned person

two phase:
●teacher phase: all the other solutions(students/learners) are updated based on the teachers’ value
●learner phase: students’ value are fed as inputs to the learner phase while learner are interacting with other learners to improve their knowledge/solution quality

uses only common control parameters such as population size and number of generation

more usable, hence used recently to solve diverse optimization problems
Camp and Farshchin, 2014
Rao and Kalyankar, 2013

Physically inspired algorithms
SA
modified version of the heuristic search algorithm: stochastic hill climbing
加上了概率接受
Goffe et al., 1994,
Skiscim and Golden, 1983

可以看得出来作者写到SI有点敷衍，哈哈。 下一章终于开始结合实际情况进行说明了。 不过相关的论文还没开始看… Hope they are not so deeply inter-dependent :=)