Local Search: Local Beam Search and Genetic Algorithms Part-7

[@phoenixrising164]

Excellent lecture . Appreciate sharing this . Envy the folks who are taught by him.

[@akshatsingh5475]

Legend

[@KUNALGUPTA05]

Crossing over was definitely not the word I thought of

[@msid9870]

yeah me too dude. i was like where is this professor taking the lecture to! XD

[@spidermanhomecoming001]

GENETIC ALGO SUMMARY: 1. Started off with K-random nodes/parents (called POPULATION). 2. Next we chose the best parents on the basis of FITNESS FUNCTION. The parent/state which had the highest pair of non attacking queens, was chosen as the best parent to mate with other parents.(NATURAL SELECTION) 3. When it came to crossing over the best parents, we randomly chose a combo of traits from each parent. And crossed it over.(CROSSING OVER) 4. Once we got a child, we just randomly mutated/changed just a single trait.(MUTATION) Thus we ended with 4 completely different states. Basically you got a new population. 5. Now you can repeat the process.

[@spidermanhomecoming001]

13:00 So u ll have a list like 84729435 12567894 47265829 67212429 35647235 So u have started off with random 5(K) states. This list is called the population. But which parents are good? How do we assess that? We need a function called the fitness function. U give it a state it will tell the number of non attacking queens. The higher non attacking queens the better it is in our N queens problem. So.... 1. Choose the 2 best parents from the K nodes. 2. These parents will produce children thru 2 steps. One is called CROSSING OVER another one is called MUTATION. Imagine the state with high probability to be like a rich n good looking man. Bcoz of this he gets to mate with more than one female to produce babies. So we gonna partner him up more than once with other states to produce babies. In the image, We randomly choose the 2nd state(with 29% probability) to mate(have sex) more than once. LMAO. But the 4th/last state 32543213 is weak....with only 11 non attacking queens n very low prob of 14%, so he doesnt get to mate. He will die a virgin. LOL. U choose one male state 32752411 to mate with 2 females 24748552 and 24415124. But the 4th state is left ignored. If u see in the 2ND PART(PARENT SELECTION) u can see the same parent 32752411 is chosen twice to mate with other 2 nodes. For the 3RD PART(CROSS OVER), we randomly choose 1st three traits(327) from the 1st parent and the balance 5 traits(48552) from the 2nd parent. And we mix it up, so we endup with a new child (32748552). And in the LAST 4TH PART(MUTATION), we randomly change a single number in the child. Its called random mutation. Therefore the child goes from (32748552) to (32748152). Thats what happened in the 1st row. But in the 2nd row you dint mutate the child. But thats ok. Coz this is RANDOM MUTATION. Just like that these 3 processes happen in 3rd and 4th row.

[@smitasmathematica.1041]

Respected Masum Sir kindly explaining the algorithm that you tell in this vedio how genetic algorithm explain multiple partners create a good quality generation next product?It is not clear in this vedio.