And 3 years after your comment he is still teaching people interesting things ! There is something amazing in that ! In my case he is even reaching to Belgium to teach me something new :P
I created a genetic algorithm to find combinations of weights and ingredients in meals that meet a person's desired nutritional criteria, both macronutrients and micronutrients. The solution was created in Go without using any library. I absorbed the concept of genetic algorithms and decided to implement something that met my exact objective. I was very happy when I noticed that the results were incredibly satisfactory. A varying number of macronutrient and micronutrient restrictions could lead to meal combinations with ingredients that are very close to what is expected. I am Brazilian and I intend to launch this feature in Brazil in the next few days. If anyone is interested in knowing details about this, please don't hesitate to respond.
Intellectually stimulating, the educator was very effective at cutting through large swaths of information summarily articulating them in ways I believe suitable for the students present. Very complex subject matter made easy and enlightening.
When comparing to my Indian classes, I'm seeing some major differences of the foreign university The instructor was very friendly I don't see the Instructor sitting during whole class He is wise enough to take the class and can handle the young chaps Students entering class without his permission They didn't sit properly before the instructor
Ohh buddy, here in the USA we watch Indian Professors on UA-cam to explain us things that our professors are not able to, so you guys have talent for teaching, too. But you are right this Professor in particular, is extraordinary at teaching things.
if programmer did not know about the golden rules of crossing, mutation and fitness, and got these 3 tricks/observations by himself, a big big big nobel prize must be granted to him!!!
He's explaining the concept from its fundamentals, that way the students can understand not only the simple concept, but how that concept was formed. It's kind of like a math teacher writing a proof for a theorem, and explaining what the logic is between each step. Sure, you can use the equation all you want, but you won't know if you've made a fundamental error unless you know the fundamentals.
In the natural world, it isn't a programmer that deserves credit, rather the genetic algorithms and the richness of the space. In the artificial world, I see no reason why the richness of the space and the ingenuity of the programmer deserve more credit than the genetic algorithms themselves. Why shouldn't an artificial environment have predispositions, perhaps even inevitability, just like evolution?
InfiniteUniverse88 Because on a simulation you want your whole population to be genius and entrepreneurs, the world is full of ordinary people, but you can't afford having a simulation that have 7billion entities, and just a few are extraordinary. Thats why he said it is naive.
Really? I got the distinct impression that he was. From his statement that we don't know how species evolved one can see he never read or understood Darwin's "Origin of Species". Later he feels compelled to ask the unrelated question of "where does the credit lie" and answer that it is with the designer. These are hallmarks of a creationist. Sorry. The natural environment is a billion times more rich in solutions than any simulated environment. He is right about these algorithms being simple (or "naive", as he put it). There is no change to the length of the genes or any possible alternate application of any gene other than what the fitness function checks for. In biology the only fitness function is "can you breed successful breeders" not caring at all what solutions one employs to that end. Anything goes... including genes that do nothing at all but get passed on with mutations and exploring the unknown space of novel parts to add to the genotype. "Don't know how species evolved?" Give me a fucking break.
RazorX53 *" "Naive" has a specific meaning in this context."* I thought I pretty much covered that pretty thoroughly in my post. Didn't I? Sorry, his ending was even more damning. Yes he does credit the richness of the solution space where damned near anything can address the challenges, But to not realize it's simply the interaction of a non-program (the incessant iterations of filtering of variants) that does the job. Why else would he need to credit a programmer rather than the inherent math of accumulating beneficial new genes?
Hey guys, I might be a bit thick here, but what does the professor mean when he says near the end of the lecture -> "We were amazed by the SPACE of solutions ... and not by the GENETIC algorithms'? Any further explanation is welcome :)
I guess, He tries to tell that algorithm is not perfect and not able to provide precise solutions every time, because it is a metaheuristic algorithm. But what GA provide is the possibilities of solutions that human can not even imagined
actually, this video is almost 3 years out of date. OpenAI's neuroevolution algorithm (run in parallel among 2000 cores) was able to solve Atari games faster than Google's DeepMind, which uses Reinforcement Learning and backpropagation or something. but basically, if you have a whole company's resources to cores, then neuroevolution is the fastest way to teach a.i. to play video games, because it's much more parallelizeable.
è interessante , come da figure semplici si possano commutare e , esponenzialmente rendere sempre piu complesse , per questioni che vanno oltre la teoria die colori , nodi e altro ecco , cosi complesso da definire un esponenziale mutuale .
Indeed. The answer, and its utility, depends on how well the question is framed. That simulation was more a 'hockey fight' than a general environment with a need for food built in.
The lecture videos for Genetic Algorithms (GA) are already been uploaded in ua-cam.com/play/PLsEIbHOtypITmujPz-TKmWsMH5eqbFgpf.html (from theoretical perpectives) ua-cam.com/video/mwXckn8up_U/v-deo.html (how to write code) Please give your valuable comments after watching the videos.
I find interesting that if you choose Pc < 0.5, then Pn would be grater than Pn-1, because 1-Pc > Pc. Does this mean that you should always choose Pc >= 0.5?
theEyE no it won't be greater just try it. let Pc be 0.3 P1 = 0.7^0 * 0.3 = 0.3 P2 = 0.7^1 * 0.3 = 0.21 P3 = 0.7^2 * 0.3 = 0.147 P4 = ... (1-Pc) will be greater than Pc, but smaller than 1 and than you multiply by pc and the result will be smaller than both factors.
Wischenbart Christian By Pn i was referring to the last probability. Let's say there are 4 individuals in your population. If you choose Pc = 0.3, you would have: P1 = 0.3 P2 = 0.21 P3 = (1-Pc)^(3-1) = 0.7^2 = 0.49 Thus P3 > P2. In the general case, let K be the number of individuals in your population, and Pc < 0.5. Pk-1 = (1-Pc)^(n-2) * Pc and Pk = (1-Pc)^(n-1) = (1-Pc)^(n-2) * (1-Pc). Because Pc < 0.5 => (1-Pc) > 0.5 => Pk>Pk-1
I assume you mean 'few' as opposed to 'little'. They're normal size. But, it's because MIT is a highly competitive school with about 11k students not 30k. This is a specialized class for undergrads and MIT is 60% grad students. There are many courses to take and many interests people pursue. MIT also wants a good student to faculty ratio.
Best greetings from Germany ! I'm a high school student in Germany and I think AI and these algorithms are very useful and interesting. In Germany the most people don't care about it today, but our politians try to move the people in these for them new direction. In the direction of self learning machines, machines who do the most job of us. For example helping doctors while they run diagonstics on their patients or do operational things... ;) Maybe It's a huge thinking forward, in the future.
There's an incomplete subtitle line here: 13:59: "So we'll just truncate anything like that at 0" Translations are locked so I can't correct it. MIT pls fix
because the next generation values are based on the original values, so they will be similar to the original values and might get stuck into a local maximum. You mutate them to try o get out of local maximum. Like imagine if all the random original values were 1, you would only get 1s in the next generation without the mutation.
Hi. I would like to use Genetic Algorithm in MATLAB to run Rotating Disc Contactor (RDC) Column data. Can u teach me how solve this problem ? Thank you for your time and consideration.
I watched your lecture with great interest. I'm teaching myself Python by coding a GA. Often, when selection and reproduction are discussed, the biological model of two parents are combined into one offspring. I have a different idea. Say you have a starting population of 200. You apply your fitness function to score each member and then the grim reaper function to kill the bottom half in terms of fitness. You have a population of 100 members. Why not combine each member with every other member? (think nested loops). 100 * 100 (crossover) produces 10,000 new members. apply a mutation function randomly against the population and against each cell in the DNA string. Then reduce the population by 99% by fitness back to the original level of 100. In effect producing the next generation from the top 1 percent of the current generation. Have you considered such an approach? Can you give me your opinion? Thank you!
My population is a 2d array. the 0 column is the genetic string, the 1 column is the fitness score of the 0 column. If your population is 200, then delete by fitness score < than the average fitness score. The remaining population (100) is the most fit of the origonal 200. The question is how to recombine the 100 to produce a new generation that improves the fitness score without losing diversity? If you recombine all 100, 100 times, that creates a new population of 10,000. Now calculate the fitness score of the top 1% and eliminate the rest. You're back to a population of 100 but that new population has a dramatically better fitness score. I'm concerned that I'm trapping the evolution in a sort of local minima / maxima sort of thing.....
That's the concept of genetic drift, is it not? where you're having a bottleneck effect occur every generation, and not using a natural selection based algorithm that would include 'inclusive fitness' and regular fitness to the number of offspring produced.
This is very helpful for me. But I have a question. What is Pc ? And how much is it. I watch the screen ,find the rank probability is 0.05. (1-Pc) equals 0.95,so 0.95^39 always more than 0.05,if Pc equals 0.05. I think I need some help.
Kinda disappointed by this lecture: 1. The lecturer said mutation is essentially hill-climbing which I agree. But he didn't explain what cross-over is and why it is important. At least he should have stressed that it was still a mystery. 2. Crediting the artificial creature program for its "rich solution space" rather than genetic algorithm without even justifying it is kinda irresponsible. Because that's a bold and non-trivial claim. 3. Yes, GA requires fine-tuning of parameters, in machine learning we have feature engineering which is doing the same thing. Isn't it naive to thinking an algorithm as general as GA would work well on all problem instances without feature engineering? There is no universal problem solving algorithm that works well for all problem instances (no free lunch theorem) Overall, I have the impression that the lecturer has prejudice against GA.
Rest in peace. But too bad he had to spread misinformation and nonsense about Evolution. The classic "we know how certain changes can develop, but not how to jump from species to species". Of course we do, we even observed it: it's just many "small changes". USA and creationism, damn...
no, a fitness of zero is not leaving any offspring after you die. a negative fitness is taking more of the genetic material that you share with others out of the world, most likely through killing/being the reason for a net loss in family members.
realize that fitness is just an arbitrary function that you set yourself, there is no fundamental "meaning" behind a fitness being positive versus negative.
Classes aren't often place where people feel that it's appropriate to laugh. I've been in classes where professors do joke and have great energy. We don't raucously cheer or anything, but we smile and chuckle quietly. And I'd like to think the professor appreciates it.
Is interesting to see how uncomfortable this professor is with the idea of evolution despite he seeing that it works and that is able to generate great solutions. Trying to bring God in an algorithm where there's no need for it. He did a good explanation of the general idea though.
wowow, good question. It should be the algorithm itself because programmer just mimick correctly the golden rule of God. Programmer did not invent anything new.
Lunasyke erm...these algorithms were inspired by darwins theory of evolution and natural selection and other biological facts? So yes, biology is just as important as computer science.
The creationist based inaccurate interjections are very unprofessional and unfortunate. I'm not saying he's not covering the subject effectively, but he is generalizing in unsubstantiated ways in fields which inspired this topic for no positive reason.
there are tons of other videos on youtube describing everything he does much faster more accurately and in a much more interactive way but i guess this is a lecture and he's just doing his job whereas most of the youtube videos on this subject were made to be online
Anti Matter Dynamite I agree that he looks like he is bored all the time. But i think he's explaining the concept of the lecture step by step very well, especially to someone who has never taken it before, or haven't understood it.
Patrick Henry Winston (February 5, 1943 - July 19, 2019). Thank you for your courses.
Ban Ma so Sad.. he seems good person..
@Brad Watson hi brad, you might be watching the wrong courses..
@Brad Watson F off, there is no god
And 3 years after your comment he is still teaching people interesting things ! There is something amazing in that ! In my case he is even reaching to Belgium to teach me something new :P
He seems a great teacher. RIP
I really like the candies, black boards and the genetic algorithm demos. Many thanks Prof. Patrick and MIT
I created a genetic algorithm to find combinations of weights and ingredients in meals that meet a person's desired nutritional criteria, both macronutrients and micronutrients.
The solution was created in Go without using any library. I absorbed the concept of genetic algorithms and decided to implement something that met my exact objective.
I was very happy when I noticed that the results were incredibly satisfactory. A varying number of macronutrient and micronutrient restrictions could lead to meal combinations with ingredients that are very close to what is expected.
I am Brazilian and I intend to launch this feature in Brazil in the next few days. If anyone is interested in knowing details about this, please don't hesitate to respond.
Hi Luis, I’m interested in knowing the details as I want to do something with genetic algorithms also in go. How do I reach you?
Excellent!!! One of the best ML courses I’ve seen. Thanks MIT for sharing this knowledge.
Intellectually stimulating, the educator was very effective at cutting through large swaths of information summarily articulating them in ways I believe suitable for the students present. Very complex subject matter made easy and enlightening.
When comparing to my Indian classes, I'm seeing some major differences of the foreign university
The instructor was very friendly
I don't see the Instructor sitting during whole class
He is wise enough to take the class and can handle the young chaps
Students entering class without his permission
They didn't sit properly before the instructor
Ohh buddy, here in the USA we watch Indian Professors on UA-cam to explain us things that our professors are not able to, so you guys have talent for teaching, too.
But you are right this Professor in particular, is extraordinary at teaching things.
Yes NPTEL is too boring to watch these are atleast intresting
@@TheDjarto Lamest joke ever. Which Indian professors do you watch, son? Speak for yourself. No-one does that, not even Indians.
@@JthElement ohh plenty of them, not a joke. I no longer have a need to watch academic material but the guy I remember is Abdul Bari
if programmer did not know about the golden rules of crossing, mutation and fitness, and got these 3 tricks/observations by himself, a big big big nobel prize must be granted to him!!!
It's just taken from evolution
Nice lecture. RIP professor Patrick Winston.
It was a clear and useful taught. 100% recommended.
Wow, this is a ridiculously convoluted way of explaining such a simple concept.
He's explaining the concept from its fundamentals, that way the students can understand not only the simple concept, but how that concept was formed. It's kind of like a math teacher writing a proof for a theorem, and explaining what the logic is between each step. Sure, you can use the equation all you want, but you won't know if you've made a fundamental error unless you know the fundamentals.
Some Random Guy Reading your comment makes me shaking my head.
profoundly interesting - I have found gold .
Thanks , for this educational contribution.
R.I.P Patrick Winston
In the natural world, it isn't a programmer that deserves credit, rather the genetic algorithms and the richness of the space. In the artificial world, I see no reason why the richness of the space and the ingenuity of the programmer deserve more credit than the genetic algorithms themselves. Why shouldn't an artificial environment have predispositions, perhaps even inevitability, just like evolution?
InfiniteUniverse88 Because on a simulation you want your whole population to be genius and entrepreneurs, the world is full of ordinary people, but you can't afford having a simulation that have 7billion entities, and just a few are extraordinary. Thats why he said it is naive.
It is really worth for spending only 47 mins to know the basic concept of Genetic Algorithm
Sarcasm? You only need to spend 10 min reading through the wiki page for Genetic Algorithm.
great lecture ... humbled by the lad on the front row with the eyesight issue
Q: "Professor Winston is a creationist."
A1: No
A2: No
A3: No
Really? I got the distinct impression that he was. From his statement that we don't know how species evolved one can see he never read or understood Darwin's "Origin of Species". Later he feels compelled to ask the unrelated question of "where does the credit lie" and answer that it is with the designer. These are hallmarks of a creationist. Sorry. The natural environment is a billion times more rich in solutions than any simulated environment. He is right about these algorithms being simple (or "naive", as he put it). There is no change to the length of the genes or any possible alternate application of any gene other than what the fitness function checks for. In biology the only fitness function is "can you breed successful breeders" not caring at all what solutions one employs to that end. Anything goes... including genes that do nothing at all but get passed on with mutations and exploring the unknown space of novel parts to add to the genotype.
"Don't know how species evolved?" Give me a fucking break.
RazorX53 *"
"Naive" has a specific meaning in this context."*
I thought I pretty much covered that pretty thoroughly in my post. Didn't I?
Sorry, his ending was even more damning. Yes he does credit the richness of the solution space where damned near anything can address the challenges, But to not realize it's simply the interaction of a non-program (the incessant iterations of filtering of variants) that does the job. Why else would he need to credit a programmer rather than the inherent math of accumulating beneficial new genes?
Unfortunately i too believe that this guy is a creationist... hope i'm wrong.
Why do you care if he is a creationist? He is a great lecturer regardless.
John L.
Obviously because he does not grok the subject.
outstanding teacher. Thanks a lot. But can anybody explain it's real life application.
en.wikipedia.org/wiki/List_of_genetic_algorithm_applications
The real life applications are too many to count.
Hey guys, I might be a bit thick here, but what does the professor mean when he says near the end of the lecture -> "We were amazed by the SPACE of solutions ... and not by the GENETIC algorithms'? Any further explanation is welcome :)
I guess, He tries to tell that algorithm is not perfect and not able to provide precise solutions every time, because it is a metaheuristic algorithm. But what GA provide is the possibilities of solutions that human can not even imagined
Thank you a lot for this course.
The problem with homogeneity on species (as opposed to individual) performance: @27:12
Many thanks for the lecture... All we educated
i wish I had a professor like this
Highly intriguing and informative.
actually, this video is almost 3 years out of date. OpenAI's neuroevolution algorithm (run in parallel among 2000 cores) was able to solve Atari games faster than Google's DeepMind, which uses Reinforcement Learning and backpropagation or something. but basically, if you have a whole company's resources to cores, then neuroevolution is the fastest way to teach a.i. to play video games, because it's much more parallelizeable.
actually it's slightly more than 3 years out of date
@@fuzzypenguino it's slightly more than 5 years out of date now
This whole video is a stealth ad for Weight Watchers International.
è interessante , come da figure semplici si possano commutare e , esponenzialmente rendere sempre piu complesse , per questioni che vanno oltre la teoria die colori , nodi e altro ecco , cosi complesso da definire un esponenziale mutuale .
Swank bro. I was able to craft some sweet bots with these algorithms. Great lesson.
Thanks for the lecture, learned so much
R.I.P professor Patrick Henry Winston
Shouldn't train them to compete for food but rather program with a need to eat, then you see if they share or work together instead of war.
Indeed. The answer, and its utility, depends on how well the question is framed. That simulation was more a 'hockey fight' than a general environment with a need for food built in.
GREAT MAN
RESPECT
great, evolution is the gold key that God gives to humans.
I wish I had gone to MIT!
Me too
Awesome ,thanks a lot.
Awesome and easier, thanks!
The lecture videos for Genetic Algorithms (GA) are already been uploaded in
ua-cam.com/play/PLsEIbHOtypITmujPz-TKmWsMH5eqbFgpf.html (from theoretical perpectives)
ua-cam.com/video/mwXckn8up_U/v-deo.html (how to write code)
Please give your valuable comments after watching the videos.
Hey guys if we remove the diversity factor at some instance i.e. generation then that generation could be perfect?????
Thanks MIT
Can anyone give me the link to the demo shown on the video
43:43 Really makes me shiver how human-like they behave. And makes me wonder if these animations were really generated by a GA.
1:30 he didn't choose the thug life, the thug life chose him
@Brad Watson ..... nani?
This feels like computer church theological seminary discussion on the computer coding theology
Jk
Awesome instructor BTW
I find interesting that if you choose Pc < 0.5, then Pn would be grater than Pn-1, because 1-Pc > Pc. Does this mean that you should always choose Pc >= 0.5?
theEyE no it won't be greater just try it.
let Pc be 0.3
P1 = 0.7^0 * 0.3 = 0.3
P2 = 0.7^1 * 0.3 = 0.21
P3 = 0.7^2 * 0.3 = 0.147
P4 = ...
(1-Pc) will be greater than Pc, but smaller than 1 and than you multiply by pc and the result will be smaller than both factors.
Wischenbart Christian By Pn i was referring to the last probability. Let's say there are 4 individuals in your population. If you choose Pc = 0.3, you would have:
P1 = 0.3
P2 = 0.21
P3 = (1-Pc)^(3-1) = 0.7^2 = 0.49
Thus P3 > P2.
In the general case, let K be the number of individuals in your population, and Pc < 0.5.
Pk-1 = (1-Pc)^(n-2) * Pc and Pk = (1-Pc)^(n-1) = (1-Pc)^(n-2) * (1-Pc). Because Pc < 0.5 => (1-Pc) > 0.5 => Pk>Pk-1
So did you know the Pc value?
love how the guy in the beginning just gives the basket but did was the first to get this chocolate xd
🤣🤣🤣🤪😎
which language and software is being used here to get the test runs?
Thanks Professor Winston, for teaching me that sharks are not just good at murdering fish, they're really good at murdering fish.
Is that a university for a masters degree? Why so little people.
I assume you mean 'few' as opposed to 'little'. They're normal size. But, it's because MIT is a highly competitive school with about 11k students not 30k. This is a specialized class for undergrads and MIT is 60% grad students. There are many courses to take and many interests people pursue. MIT also wants a good student to faculty ratio.
Kenny is the true hero.
Best greetings from Germany !
I'm a high school student in Germany and
I think AI and these algorithms are very useful and interesting.
In Germany the most people don't care about it today, but our politians try to move the people in these for them new direction.
In the direction of self learning machines, machines who do the most job of us.
For example helping doctors while they run diagonstics on their patients or do operational things... ;)
Maybe It's a huge thinking forward, in the future.
There's an incomplete subtitle line here:
13:59: "So we'll just truncate anything like that at 0"
Translations are locked so I can't correct it. MIT pls fix
Thanks for your note! We've update the caption.
How to use Genetic Algorithm in MATLAB SPLIT RING resonator design
Why does the professor have a list of pictures (i guess of students faces) listed at 18:27?
So that he can see their names. He sometimes calls the students by their names in the lecture hall.
Great lecture!
Why would i need to change the values (mutate them) to random values if i the original values were created randomly too?
because the next generation values are based on the original values, so they will be similar to the original values and might get stuck into a local maximum. You mutate them to try o get out of local maximum. Like imagine if all the random original values were 1, you would only get 1s in the next generation without the mutation.
@@fsy14 yeah ,i don't know what i was thinking
Is this filmed with an auto-tracking PTZ camera?
I think they got Spielberg in to do it.
x xenocide :D
the tutor needs to do some fitness, I was about to sleep listening his suffer breathes
+Morphius he believes in chocolates as good soft drug before lectures and quizes.. can you blame him? lol.
Hi. I would like to use Genetic Algorithm in MATLAB to run Rotating Disc Contactor (RDC) Column data. Can u teach me how solve this problem ?
Thank you for your time and consideration.
I watched your lecture with great interest. I'm teaching myself Python by coding a GA. Often, when selection and reproduction are discussed, the biological model of two parents are combined into one offspring. I have a different idea. Say you have a starting population of 200. You apply your fitness function to score each member and then the grim reaper function to kill the bottom half in terms of fitness. You have a population of 100 members. Why not combine each member with every other member? (think nested loops). 100 * 100 (crossover) produces 10,000 new members. apply a mutation function randomly against the population and against each cell in the DNA string. Then reduce the population by 99% by fitness back to the original level of 100. In effect producing the next generation from the top 1 percent of the current generation. Have you considered such an approach? Can you give me your opinion? Thank you!
John David Deatherage How are you sure you won't take out the other top ones when you reduce the population?Newbie here
My population is a 2d array. the 0 column is the genetic string, the 1 column is the fitness score of the 0 column. If your population is 200, then delete by fitness score < than the average fitness score. The remaining population (100) is the most fit of the origonal 200. The question is how to recombine the 100 to produce a new generation that improves the fitness score without losing diversity? If you recombine all 100, 100 times, that creates a new population of 10,000. Now calculate the fitness score of the top 1% and eliminate the rest. You're back to a population of 100 but that new population has a dramatically better fitness score. I'm concerned that I'm trapping the evolution in a sort of local minima / maxima sort of thing.....
That's the concept of genetic drift, is it not? where you're having a bottleneck effect occur every generation, and not using a natural selection based algorithm that would include 'inclusive fitness' and regular fitness to the number of offspring produced.
Well real populations don't breed across the population. And it would be too time consuming.
Thank you
what about the parameters in the video?I have tried many times ,but can't find the best parameters,please help！ thanks
hello,Yu Wang.I don't know it.But I want know the parameters Pc. Can you help me if you find the Pc value please? Thanks u in advance.
This is very helpful for me. But I have a question. What is Pc ? And how much is it. I watch the screen ,find the rank probability is 0.05. (1-Pc) equals 0.95,so 0.95^39 always more than 0.05,if Pc equals 0.05. I think I need some help.
It's a freely chosen probability, 0 < Pc
Does the basis of designer babies use genetic algorithms to calculate phenotypes and outcomes thereof?
6:38 Does anyone see the "thing" in green shirt in the front row? I got freaked out for a second...
Kinda disappointed by this lecture:
1. The lecturer said mutation is essentially hill-climbing which I agree. But he didn't explain what cross-over is and why it is important. At least he should have stressed that it was still a mystery.
2. Crediting the artificial creature program for its "rich solution space" rather than genetic algorithm without even justifying it is kinda irresponsible. Because that's a bold and non-trivial claim.
3. Yes, GA requires fine-tuning of parameters, in machine learning we have feature engineering which is doing the same thing. Isn't it naive to thinking an algorithm as general as GA would work well on all problem instances without feature engineering? There is no universal problem solving algorithm that works well for all problem instances (no free lunch theorem)
Overall, I have the impression that the lecturer has prejudice against GA.
Was getting dizzy. He walks around a lot. a bit distracting.
good!
Thanks Sir, this very usefull for my insomnia :v
Rest in peace. But too bad he had to spread misinformation and nonsense about Evolution. The classic "we know how certain changes can develop, but not how to jump from species to species". Of course we do, we even observed it: it's just many "small changes". USA and creationism, damn...
What program is he using?
You can find lot of gui based genetic algorithms on UA-cam
please , give me some link of this software.
Hasanul Islam - Why not create your own?
UNHEALTHY, WILL ALL THE 'GENIUSES' AT MIT ------>.>>>
44:58 he looks like an angry gorila, mad because he couldnt get the food
lol 13:50. negative fitness. Is that like dying before being born? :)
J O most of the time, negative fitness means: "you did worst than doing nothing"
J O No, it is dying before reprodutive age.
no, a fitness of zero is not leaving any offspring after you die. a negative fitness is taking more of the genetic material that you share with others out of the world, most likely through killing/being the reason for a net loss in family members.
realize that fitness is just an arbitrary function that you set yourself, there is no fundamental "meaning" behind a fitness being positive versus negative.
Once again, it's an algorithm based on hill climbing. Some of the hill does have that deep valley that might be negative.
The awkwardness of the crowd made me not watch the video. Guys, you need to laugh sometimes.
Classes aren't often place where people feel that it's appropriate to laugh. I've been in classes where professors do joke and have great energy. We don't raucously cheer or anything, but we smile and chuckle quietly. And I'd like to think the professor appreciates it.
People laugh towards the end of the video ;)
hahahah, terribly unfortunate! turns out to be the lucky thing to save us :D
Is interesting to see how uncomfortable this professor is with the idea of evolution despite he seeing that it works and that is able to generate great solutions. Trying to bring God in an algorithm where there's no need for it. He did a good explanation of the general idea though.
I thought the title meant 13 different genetic algorithms.
LOOOL legal drug giving for free wish our Prof is as cool as him!
@40:40
Im glad i did not pay for that. but thanks anyway.
Well, there we go. I can at least get one mark on an MIT exam. He's definitely. a creationist.
2:50
wowow, good question. It should be the algorithm itself because programmer just mimick correctly the golden rule of God. Programmer did not invent anything new.
cooooooool.,...
he has to lose some weight
hahah, giao trinh hoc tieng anh hay nhut nhut the gioi thien ha vu tru day ne :D deo phai toefl hay ielts :D:D
Real biology has many more variables of course.
+Galaxy Spirals really? no way!
Galaxy Spirals - This is not biology. this is algorithms programmed to evolve.
Lunasyke erm...these algorithms were inspired by darwins theory of evolution and natural selection and other biological facts? So yes, biology is just as important as computer science.
The creationist based inaccurate interjections are very unprofessional and unfortunate. I'm not saying he's not covering the subject effectively, but he is generalizing in unsubstantiated ways in fields which inspired this topic for no positive reason.
First two minutes are yummy :))
Boring as hell
this guy is so boring.... and he chooses to present the material in a very non intuitive way
How would you present it?
there are tons of other videos on youtube describing everything he does much faster more accurately and in a much more interactive way
but i guess this is a lecture and he's just doing his job whereas most of the youtube videos on this subject were made to be online
link please?
Anti Matter Dynamite I agree that he looks like he is bored all the time. But i think he's explaining the concept of the lecture step by step very well, especially to someone who has never taken it before, or haven't understood it.
i guess he's appealing to an audience that has no idea what he's talking about then
3:12