Understanding the Cuckoo Search Algorithm: Principles and Implementation

Hello, In this video, I will discuss Cuckoo  Search Algorithm. Topics that are covered in   this video. What is cuckoo search algorithm?  Basic principles that are used for simplicity   and then we will learn How we can implement this  algorithm step by step? What is levy flight? levy  

distribution and how a cuckoo bird can generate  new solution? Its mathematical model and the Aim   of this algorithm? Then how we can compare the  fitness value of the cuckoo bird with the host   bird and then we have the parameters that you  can use for the implementation of this algorithm  

and this parameter are sufficient for all type  of optimization problems so first we will start   this with what is cuckoo search algorithm. It is  a meta heuristic algorithm that is inspired by   some cuckoo species link their egg in the nest of  other species bird that increase their population  

and their survival here you can see the example  of cuckoo bird egg in the other species nest   and you can see here um other  species were feeding cuckoo.   So in this algorithm we have two bird species  cuckoo bird and other bird we are denoting here  

other bird with host bird so we have cuckoo  and host bird you know the behavior of cuckoo   bird they do not have any nest they lay  their egg in the nest of other species   so here we have a nest for host bird and cuckoo  bird will lay the egg in the nest of host bird  

you know when the cuckoo will lay its egg in  the nest of host bird there is a possibility   of discovery of cuckoo egg by the host bird so  what happened if host bird discovered cuckoo's egg   so for this we will denote this  possibility with the probability that is  

in the range zero to one that is here p a is the  discovery of alien egg that is the cuckoo's flag   so once the egg is discovered by host word  then it can take two accents either it can   throw the egg away or it can build new nest so  you can see here if the host were discovered  

cuckoo egg the host bird can throw away the egg or it can abandon the nest and build the new one   so the three rules for the simplicity of this  algorithm are here first is each cuckoo lay   only one egg at a time and places it in a  randomly selected nest done second rule is  

best nest with high quality of egg will carry  over to the next generation third rule is the number of available host nest is fixed and  once the egg is discovered by host nest it can   throw away it or build new egg and there is  a 10% possibility of discovery of cuckoo egg.  

now we will try to convert this into mathematical  model first rule is each cuckoo lay only one egg   at a time and place it in a randomly selected  nest done here each egg represents a solution   that is stored in the nest and the artificial  cuckoo can lay only one egg at a time  

so the egg in the nest is a set of solution  and cocoa wag is the new solution done   second rule is best nest with high  quality of egg will carry over to the   next, generation here high quality egg means best  best solution here optimal value it means the egg  

that are more similar to the host word act and  have opportunity to develop a more mature cuckoo   and that is the new generation here and in order  to calculate the fitness value the new and the   better solution will be replaced with less fit  solution that's how we will compute the fitness  

value third rule is that is a number of host  nest is fixed and once the exit is covered by   host word it can either killed the egg that  is through away or it can build the new nest   and the probability is due to one a number  of host nest is fixed that is the population  

and host bird discover the cuckoo.. that is the  worst solution away from the optimal value   this is the algorithm for the cuckoo search.  let's see this with the flowchart first we   will initialize the parameter for cuckoo search..  after that we will initialize the population of  

host nest and keep in mind the population of host  nest is fixed check the condition if it is true   then generate a new solution for cuckoo  using lavish flight evaluate its fitness   after that choose the nest randomly and then  compare the fitness of the cuckoo with the host  

nest if the fitness of the cuckoo  is greater then we will replace it   and then next step so let's see this with first we  will initialize the parameters the simplicity of   this algorithm is here we use only two parameter  that is number of host nest denoted with n  

probability of discovery of cuckoo egg  is p(a) that is in the range of 0 to 1 and the probability of discovery of alien  egg is 0.25 maximum number of iteration 1000   this parameter are sufficient for a lot of  optimization problems and here you can see

now according to first rule each cuckoo  can lay only one egg at a time and you can   place it in a randomly selected nest  the question is here how cuckoo will   fly and how it will place the  egg in the nest. In the real life

bird can fly automatically you  know they are capable to do so   but in case of.... here in this case here  we will use mathematical model to   so this behavior so for this first we  will understand the basic fundamentals  

it is a metaheuristic algorithm that is  mimicking cuckoos. done.. so here in order to show the   cuckoos flying their behavior that is  generating the new solution we will use here   Levy's flight that is you can see the equation  here we will use this equation to generate the  

new solution by cuckoo here the value of i is  1 to n the alpha is the step size this is the   Levy's is exponent and the value is 1.5 and  this operator means entry wise multiplication   and this equation is for random walk that  is used by cuckoo bird to generate the new  

solution in the host nest here  Levy's flight is the walk it is done you can see here this is the random walk generated  by this equation and that cuckoo bird will use to   reach the host nest and will place the egg so  first we will understand the walk and the run  

in order to understand levy's flight first we will  understand walk and run you know walk means a move   at a regular phase and running is faster than  walk here so it's all about movement right so   in case of a walk it is we are taking smaller step  to reach the destination and in case of running we  

are taking continuous steps to reach destination  and here user is taking continuous tap with speed   let's see with an example so this is the  person and here we have the destination   first the person is walking suppose so it  will take step little steps one step two

so during walk a person will just  take steps to reach the destination   we are doing this to understand levy's flight  so you know this is the random walk done by   a person to reach the destination another way it  can run done now let's see in case of cuckoo here  

we are using levy's light to show this behavior by  this world you know for we are using this equation   here to show this behavior where we have cuckoo  and this is the destination that is the whole   host bird nest and cuckoo will use here levy's  flight that is the random walk used by this bird

here in this equation you can see  this is the new solution i is the you know our value of i is 1  to n t is the current iteration   then this is the current location of the  cuckoo this is the step size multiplication  

operator and revise multiplication and  lab is exponent now we will see this   algorithm again this algorithm is nothing  but the simulation of random walk random walk   search process that is used by cuckoo where's the  cuckoo is searching for a suitable nest for laying  

egg so what is this what is cuckoo random walk  search process used by a word cuckoo in order to   search the suitable nest for laying egg here the bird is  using best search strategy that is levy's flight   okay levy flight and levy's distribution using  levy's flight we can ... we can certainly obtain  

the larger step size and using levy's distribution  we it is a series of steps smaller steps that we   will use in the starting of the algorithm so you  can see here this is the levy's flight used here   for the cuckoo bird to generate the new solution  new solution means egg in the host nest and here  

levy flight provide random walk to the  cuckoo random step size is denoted by   levy's distribution that is this okay now let's  see how google search first we will initialize   the population of host nest after that we will  calculate the fitness value for each host nest  

after that we will generate the new solution for  cuckoo by using levy's flight after that we will   update the its location until we reach the best  point.. suppose we have here a number of host nest   and this is the random nest selected by  cuckoo for laying egg now here is situation  

this is the position of cuckoo  bird and this is the location of   destination here the word want to reach so how  the bird will travel this distance for this we   are using this equation that is the levy's flight  and that and the levy's distribution so how we  

will compute this here r(i) is the cuckoo search  radius that is the largest step cuckoo can take   and how it will it will come to know  that it is the destination when the   gold that is the host nest is within the radius of  the cuckoo within the radius of this circle then  

cuckoo will directly will fly to the destination  in a straight line that is the levy's flight now you can see here as before in the  before slide i show you the levy's walk   that is the motion of the

so once it reach... once the goal is in the  radius it will move in a straight line done.   now here you can see bird is now reached  its destination here the bird will lay   egg now next case is according to rule  number three now the cuckoo's egg is in  

the host nest now there is a possibility of  detection of the cuckoo's egg by host bird   and we will denote this with the probability of  zero to one so once the it is once the egg is   detected by host word it can throw away  the egg that is a worst case and another  

option that it can build new nest near  the old one here you can see in order to   represent this situation mathematically we  will use this equation that is the new net   new nest will be built near the old ribbon now  here you can see the rule number two according  

to rule number two the best nest with better  quality of ag will move to the next generation   so here they will talk about what is the  next generation and how the egg will move to   next generation for the simplicity  here again each nest has only one egg  

suppose the bird is here you know using  levy's flight the word is now here   it is laying its egg in the host nest so done  now once the egg is in the hose nest there is a   again probability of egg detected or egg is not  detected suppose this is the cuckoo bird egg  

and the egg is not detected by host bird here the  case is cuckoo egg is not detected by host bird   and one important thing is here cuckoo egg hatched  earlier than the host bird egg so you know cuckoo   egg is hatched then this will move to the next  generation second case is that is the worst case  

that is the ag is detected by host bird so in  this case cuckoo egg is killed that is throw away now the aim of this cuckoo search algorithm is  to use new and battery solution that is cuckoo   to replace the less good solution in the  nest to achieve the optimal solution if the  

fitness value of the cuckoo egg is greater  than the fitness value of the host egg   then we will replace the host egg with cuckoo and  increment as you can see here check the condition   if the witness value of the cuckoo is greater  than host then replace the egg if the fitness  

value of cuckoo is less than the host bird this is  the worst case then than cuckoo's egg thrown away   from the nest and then we will generate a new  solution using levi's flight again so what is   fitness fitness is also fitness function is also  you know known as evaluation function objective  

function and we use them for the optimization  and here we determine how fit a solution is   in the cuckoo search worst nest are replaced with  new solution that are the better solution than   older one and the current solution is ranked  at each iteration we will rank the solution  

until we met the condition and the  output you can see the vast nest   so this is all about cuckoo source you can see  here you can see the parameters that are used here   and these parameter values are sufficient for  maximum of division problem here n is a population  

that is the nest host nest you can take the value  25 reality of finding egg by host bird is 0.25   and dimension is 2 current iteration we will  take the counter 0 maximum is 1000 step size is 2   so this is the this is affecting the step size  during random walk and this is the step size and  

this is the levy's exponent and This algorithm is  used for the training of the neural network with   batter performance for solving nurse scheduling  problem, and in the travelling salesman problem.   So that's all about cuckoo search algorithm if  you have any question you can still comment below  

and thanks for watching this video :)