Meta Learning

Video quality	The size	Download

Information Meta Learning

Title	:	Meta Learning
Lasting	:	10.18
Date of publication	:
Views	:	61 rb

Question: How much weed did you smoke before making that intro?brbrAnswer: all the weed
Comment from : Harry S Human

So if AI meant to be a scientific glossary have idiotic nonsensical answersbrbrbr How many other AIs give crap answers as they treat it like fact
Comment from : Blazeww4207

What software you use
Comment from : Internet Drivendo

Youtube suggests me this But god, this guy doesn't know how to explain meta-learning but digress to something unrelated
Comment from : Anh Tong

Nice video
Comment from : Manish Choudhary

Ends with "I have to go find a gradient" So much meta in this 😀brbrBecause he's an IA (intelligent agent) optimised by evolutionary processes
Comment from : Dheeraj Bhaskar

neural networks are the future, the true definition of intelligence
Comment from : Rfyo Rfyo

After watching the first 8 seconds of this video I was admitted to a hospital
Comment from : Odin Games

btw, Siraj has always been pushing himself hard to learn first so that he therefore can make video to enable others to also push them to continuous learning Jesus evolution is so difficult and and it becomes a self-torturing
Comment from : Francis Kwan

Human Evolution is learning that human can enable AI to mimic human evolution to learn to revolutionarily learn brbrso , When will AI , some day in the future, learns to enable Human to learn even more revolutionarily so that Human can mimic the supernatural self-made evolution of AI ? brbrwhat's a complex world with human and human-made artificial human
Comment from : Francis Kwan

what is actually meta learning? how is it diff from evolutionary algos ?
Comment from : Gowtham Krishna NS

Your code doesn't seem to be parallelized to train multiple models in a given generation simultaneously
Comment from : Eric Glynn

Hello siraj u look more fit & young Can you pls share me your diet plan?
Comment from : Alexio

jurgen schmidhuber proposed this already
Comment from : DistortedV12

The only click that i am sure of not regretting later is the click on your channel videosYou are amazing bro
Comment from : John Hammer

5:48 I thought Lamarck's research was discreditedbrbrbrNot hate
Comment from : Matt T

So the NAS(Neural Architecture Searching) is meta-learning? What about one shot learning? Im kinda confused of those concepts
Comment from : Alex J

youtube/xat1GVnl8-k it was the Bloodhound Gang, not Emenim who said we aint nothin but mammals
Comment from : Gary Butler

This is very very good stuff! Bite-sized and full of useful information Thanks alot Suraj for the effort!🙌👍
Comment from : Willy Kitheka

How can I say thanks to you Siraj Love you bro for giving this kind of well worthy knowledge God bless you Long live my dear
Comment from : Ramesh Maddali

So basically in the future AI will be made by AI😂😂 It truly is the end of all jobs Happy holidays human race 😪😴
Comment from : Silicon Messiah

I’ve done a lot of experimentation with neuroevolutionary algorithms, but so far they have seemed very inefficient It seems that for many tasks, traditional learning methods allow for quicker training because there are fewer types of parameters to optimize
Comment from : DiamondGuy 36

So you are killing millions of sentient bots in the hope of getting the best one Kicking robots looks like a good idea now
Comment from : Reverend BlueJeans

thank you for evolutionary algorithm video
Comment from : Larry tesla spacex boring Lawrence

What is Machine Learning? jargon, jargon, jargon(runnin out of words? create'em) :P
Comment from : Saurabh Sharma

Is AutoKeras any good?
Comment from : Josh Smithsonian

Member represented by a blockchain Challenge Accepted !😗
Comment from : Meryll Essig

def to_learn(data):br return to_learn(data)
Comment from : WolfRage

Thank you
Comment from : Max Musterman

I had hopes this video would show me the Overwatch META 😅
Comment from : Marcelo Pinheiro

sir, can you make a detailed video about transpose convolution aka deconvolution
Comment from : Gurpreet Singh

Nice I was thinking Since you can program a programming language in ASM Could ASM be a choice for the genetic algorithm to choose from A way to create *any code to create any model()
Comment from : Per Lindholm

A ton of processing power is the key
Comment from : Vlad K

machotelli xD
Comment from : Maxi Tube

Hi Dude Can you please make other videos about Automated Machine Learning
Comment from : DZsam

Wao
Comment from : Rakesh Mallick

Why use an evolutionary algorithm to find optimal hyperparameters instead of using a bayesian optimization algorithm? Is it for when your function of hyperparameters is not smooth due to discrete values?
Comment from : Josh Smithsonian

Thanks for finally adopting my ideas into practice, THANK YOU
Comment from : Yordan D

Dont forget to use the tournament of 4!
Comment from : Jurate Paulauskiene

Wow Siraj, loved the infographics it make learning easer (Where are the memes ?loved those too)
Comment from : Froyo Rex

you are talking to morons; what's the difference between that and talking to AI?
Comment from : L Kd

more videos more please
Comment from : mohammad mohammadian

Inter-life learning - shows picture of Lamarckian "evolution" But the idea was correct
Comment from : Ferroneoboron

Great video, interesting topic and papers!
Comment from : Academic Connor Shorten

How do you save an alreadyy trained algorithm?
Comment from : THEGUTYLOCO

How make good ai? Need girlfriend
Comment from : Dave Youtube

You look really cute Sorry but I find you irresistible (PS I am a guy )
Comment from : अनासक्त

My attempt at a design for GAI uses meta learning to optimize the scenarios the network is able to handlebr steemitcom/machinelearning/@kharn/a-proposed-foundation-for-gai
Comment from : imkharn

We need more compute power We need to make artificial neural networks out of synthetic proteins to make "smart goo" A teaspoon will be enough to defeat all the StarCraft 2 players in the world at once Mmm mmm mmmmmm, delicious neural network goo
Comment from : Software Is Better Than Magic

Evolving AI's is definitely the way forward; but we have to be careful because this is also how you get to skynet
Comment from : JLK

This is the beginning of SkyNet
Comment from : Yattish Ramhorry

I love you Siraj
Comment from : sarmad pervez

Best example of evolution at 6:26 ;-)
Comment from : Shobhit Sundriyal

Hey Siraj Thanks for sharing your knowledge with world Is there any video on reinforcement learning right from basics, please share video link Thanks in advance
Comment from : muralidhar A

Earlier this year, Fei-Fei Li published a paper without using reinforcement learning or evolutionary computation on meta learning and achieved state-of-the art Their technique only need one fifth of the RL methods, not too sure about EC though
Comment from : Rui Wang

Awesome, i saw something similar in the nature of code book
Comment from : Esneyder palacios

This is literally awesome than traditional neural networks strategy Loved this video thanks for this great information :)) Cheers
Comment from : VultureBeast

There is a simpler algorith no need of evolution
Comment from : SaveAsss

does this solve the general purpose ai problem?!
Comment from : Vaibhav B V

Great vid!brI think it would be worth mentioning the additional complexity that comes as a cost of using meta learningbrbrCan an average joe go work on meta learning that would produce cutting edge result?brbrYou would probably need a lot of gpu compute time (exponentially more than regular training) which might make this process only accessible to well funded developers, is this the case?
Comment from : daniel f

Hey siraj, can we determine whether an image is deep fake or not using Machine learning? Its pretty troublesomebrBTW like your videos
Comment from : The Computer Engineering Channel

your graphic also evolved
Comment from : ravi shankar

This is fire
Comment from : igor

NEAT exist since 2002
Comment from : Andre Marasca

5:16 killed me 😂
Comment from : Jesse Okeya

the visuals are getting better and better Amazing
Comment from : Dr Constantinos

I actually love you
Comment from : Ryan Connolly

can we do ga's in tensorflow???
Comment from : Kislay Kunal

learnception
Comment from : Loukas K

Cool animation
Comment from : 楊皓翔

I love your video
Comment from : Nova FF

in a post a couple of months ago, i said that it would be crystal clear that someone would say that alle the problems would be fixed thx to genetic neural nets ups, sorry ' genetic deep learning', ' genetic neural , that was 20 years ago'!br( of course, tabu search will be implemeted as well, that's just a question of time)brlife is a cycle!brand thgs are not so easy
Comment from : uncle max

Is the concept same as Genetic Algorithm ?
Comment from : Techguy

Hello world it's Yash
Comment from : darksoul

It seems that the neural network evelution algorithm needs lots lots lots of HW resources, we couldn't freely use it by our personal resources
Comment from : 钟钟

Siraj, do you think it is possible to find a Machine Learning job in the UK as a graduate with almost no experience? Thanks, I only recently subbed but you content is captivating and helpful at the same time Keep it up!
Comment from : Rado