Why Deep Learning Now? | AI Revolution Documentary
Вставка
- Опубліковано 24 тра 2024
- Learn more:
developer.movidius.com/
software.intel.com/en-us/ai-a...
/ intelswnetwork
Subscribe here: goo.gl/9FS8uF
Become a Patron!: / coldfusion_tv
CF Bitcoin address: 13SjyCXPB9o3iN4LitYQ2wYKeqYTShPub8
Hi, welcome to ColdFusion (formerly known as ColdfusTion).
Experience the cutting edge of the world around us in a fun relaxed atmosphere.
Sources:
qz.com/1034972/the-data-that-...
usblogs.pwc.com/emerging-techn...
/ keeping-up-with-the-gans
Dancing AI: • Everybody Dance Now
arxiv.org/abs/1808.07371
Wifi signal mapping:
rfpose.csail.mit.edu/
Motion transfer Animals in video games:
homepages.inf.ed.ac.uk/tkomura...
Neural Network Arm:
• This Robot Learned To ...
vpg.cs.princeton.edu/
Nvidia super slow motion:
people.cs.umass.edu/~hzjiang/...
• Research at NVIDIA: Tr...
Shout out to 2 minute papers who covers the latest in AI research. Check him out here:
/ @twominutepapers
//Soundtrack//
*coming soon*
» ColdFusion Book "New Thinking" |
newthinkingbook.squarespace.c...
» Facebook | / coldfusiontv
» My music | burnwater.bandcamp.com or
» / burnwater
» / coldfusion_tv
» Collection of music used in videos: • ColdFusion's 2 Hour Me...
Producer: Dagogo Altraide
» Twitter | @ColdFusion_TV - Наука та технологія
This question has been on my mind for the better part of the past year.
Thanks so much for covering this topic in depth bro! ColdFusion never fails to deliver
Deep Learning has been around since the 70's. The why now is access to cheap storage and memory. Deep learning requires vast amounts of storage. In the year 1998 I designed a 2 TeraByte storage system for 1 million dollars. Today that same storage can be had for $20.
@@laneromel5667 couldn't of said it better. brilliant.
@Ali Neeley lying scam artist POS.
google 2tb hard drive bulk, either new desktop, or used enterprise
this is one of the few channels which makes my time worth watching any video.
i just love when this guy talks about deeplearning and AI ,hopefully this video will talk about AI helping cancer treatement
imagine if you were dismantled because the robot couldn't pick you up
reminded me of this lol
ua-cam.com/video/mrXfh4hENKs/v-deo.html
You're more likely to dismantle a robot because you couldn't pick it up before it could anyway. lol 😂
@@abdlhmdx Maybe right now, but that looks like a possible dystopian future we're headed towards.
@@someone-yw2qw nice :)
hehehe robot be like 'i did good pappa, i did good right!'
intro soundrack :
Campbell & Couzare - Long Way (ft. Cozy)
"computers aren't as intuitive as humans and need more examples". While there is some truth to that; the cat recognition example is not specifically a good example of that. Humans also need "thousands of pictures" in a sense, but we get them in a more streamlined manner: watching a cat in real life streams massive volumes of data at a child's brain. Humans brains swim in massive amounts of a real-world data. As human children get older, they can recognize new animals with less data, but the foundation of all that required mass data. (As always, of course, one must be careful not to anthropomorphize too much when comparing humans and AI. But we humans also require mass data. We just get it a lot easier.)
Or, like in "..the small children can repeat words after 2-3 times", the brain makes/simulate a multitude of variants but speak only the best result, like a GAN (general adversarial netw.) do...
Ok so are you implying that computers can ONLY be fed static images of cats, they can't possibly be given videos?
They can; and I have seen projects where video is used for source data..Unfortunately, most video is given lossy compression so it can be a misleading data source (depending on the goal.) But uncompressed video (or video with a lossless compression) can be a great data source.The efforts at "self-driving" AIs use video, for example.
I know some of the "self-driving" AI attempts use a neural net at the camera itself. The camera NN "interprets" patterns seen in the image. It then sends those interpreted patterns to the central AI rather than the images themselves. The central AI doesn't actually "see" the road in a direct sense. That solves part of the data throughput problem. Uncompressed 4K video uses 1.2 Gbps. Ironically, the human brain appears to use a similar process. The first layer of visual processing is handled by the occipital lobe at the back of the brain.
yes but currently we don't have a program that can take two photos of an animal and based of the calculations it made to recognize the cat or other first animal it saw and recognize that animal or object from anther angle, different lighting or pose of the object. It is possible and will definitely one day become a reality but programming hasn't come that far just yet.
This guy went from 200 k to 1m good luck my dude ;)
SaidPower2 I think he went from 0 to where he is at.
Bro as i sayed before you don't need to change a thing! Making videos like yoz needs alot of research,reading...learning! I personally appriciate your work bro, Love the way you just explain things in detail...watching your videos is like being in a very fun class,learning n knowing stuffs the best way a a person can learn. Am from Tanzania,Africa. Keep it up brother!
Engineers with the mind of an artist are so rare, and you just happen to be one of them bro. You're gifted man, don't take it for granted.
I think engineers are creative too, like inventors.
That's why I specifically used the phrase "mind of an artist". I am also an Engineer, and I have the mind of an artist. I am not implying that Engineers are not creative, I meant that most of them cannot generate artwork.
Thank you for another amazing video. I am always very impressed with the quality of the videos you produce. Coldfusion is my favorite UA-cam channel by far!
Your content is so good. Well presented, edited, researched, written, pic selection... Thank you sir!
I wastalking about the question just recently with my collegue, nice you made a video about it :) love the channel
Awesome video, very useful overview of deep learning. I also enjoyed hearing Mogwai in the background!
I think it is important to point out that you completely failed to mention the advances in hardware and software used to train neural networks. Executing them is a simple task in comparison to actually training them on the enormous data sets they require, and in just a couple of years, a task that would have taken years to train for can now be trained for in a couple of weeks. This mostly due to Nvidias efforts to produce some insane hardware, and creating special GPU based processors for processing nerural networks.
In addition to this, something that has recently happened, which will be a game changer, is that game engines like Unity now has support for creating super realtime simulations with scenes with visuals close enough to the real world that self driving cars can be trained there, instead of relying on data from the real world. This means that as long as a realistic simulation can be made, one does not need nearly as much data from the real world anymore, which is huge!
The video is sponsored by Intel, see at 0:42. Not-so-subtle product placement at 3:47 - 3:55 and all throughout. This is essentially an infomercial for Intel. No wonder no mention of Nvidia.
The Huawei Mate 20 with its Kirin 980 and second gen neural engine can shoot video with "portrait mode" or black and white video with only a person or an object being colored. All in real time and all happening locally without connecting to the cloud!
Great video as always Dagogo!
What a great video. You are making such a good work. The content is very good, so Profesional. I’m following you since 3 years ago and ever the videos are going better and better. Congrats.
Best technology channel on the planet today
All the other "Tech" channels mostly crib amount smartphones and gadgets and all that crap!
This man makes videos on actual TECHNOLOGY!
MUCH love and respect, fam!
💥❤️
Great video !!! So interesting & thought provoking. I wish this was brought up in main media. It's like we are on the verge of something game changing & the process of it all is just gonna pass us by.
YES, I love your coverage on this field of science. Please do more AI /Robot updates as they advance. Thank you for your awesome content
Love your deep learning videos.
What are the best ways for keeping up to date with deep learning advancements? besides this channel :)
Thank you!
Amazing content, Amazing music. Dude release an album already!
I actually never buy books without any reviews available.
But your video's are always TOP quality and I love the stuff you talk about!
So I pre-ordered your book just now! Because I am really curious of the content and want to support the channel!!
Your voice is so soothing, I can't even distract for few seconds while watching your videos.
As usual. Great content. It’s always exciting when you upload something.
Watching Coldfusion on my 4k TV is godlike.
Definitely one of your best videos fam!
Thanks for the knowledge.
You have a great taste in music. I noticed some Mogwai playing at about 7:30 :). Great videos too. Keep up the good work.
I love your intros and outros!! So professional!
Im an English learner, and l love your videos very much. I'd appreciated that if you could put your subtitles on the video, i would understand the meaning better and exactly. Keep going Dagogo! Love you!
Waking up in Australia, your, and now my nation, I saw your video. I immediately tapped the shit out of it.
A coldfusion video is a cause for celebration!
Man this video is totaly knowledge giver, truly saying I become more clear about AI, machine learning and deep learning.....!
Your Chanel is amazingly awesome 😄👍👍👍
It gives me new tech update and information about the letest technology (hardware and software) keep it up.....
I'll be waiting for the next topic of next video!!
Till then, have a good one 😉😉😊
This probably the best channel I've subscribed since the first day I signed up to youtube
Great work. Love your content and style. Thanks!
The song @ 6:57 is Take me somewhere nice by Mogwai
Absolutely amazing. I can't wait to see the advancements made in just ten years from now.
A suggestion for a deep learning application:
Develop space robots that can make themselves, and eventually build stuff in space like Elysium style stuff, drilling out asteroids etc. and making space stations inside (and have the asteroid revolving for artificial "gravity").
This would be only possible if we could like create a robot "space economy" (separate to the usual world economy) and maybe it will eventually cost us nothing! Otherwise (with our traditional system) it would be cost prohibitive. And also this would be only possible with AI deep learning (and of course after the mammoth task of starting it up). But imagine being able to build anything out there for nothing, space hotels, big telescopes, space labs etc. and habitats to keep us surviving and multiplying without any worries.
I'm obsessed with this idea and that's why I clicked on this video. Good video, and it only encourages me more!
I have never seen a youtuber more talented than you. Wonderful videos.
Nice to see you got a big fat sponsor (Intel). I hope they're paying you well :)
Thanks for the great explanation.
Great buddy...👍nice explanation
Amazing content..love you videos. Keep up the good work
Some high quality video here😍 Love it!👌
Thanks for this. Part of mogwai in soundtrack melt my heart
i like Cold Fusions background music , it make the content more exciting
I love the songs you use at the beginning of most of your videos! Wish I knew the names.
*AI is a tool. A magic tool.*
Whoever learns how to use AI as a tool in their industry, will rule that industry.
Or maybe we are witnessing the birth of the wonderful AI friend, that it 1000X smarter than us and can help us to figure out how to live a happy and peaceful life on earth, instead of fighting conflicts and messing up the climate all the time :-)
Or it will kill us all cause that will be the best way to obtain peace :)
We are its biological bootloader.
Heaven or Hell, seems will find out where this ride is going pretty soon
"task: Eliminate humans suffering. Calculating...Human suffering caused by humans. Calculating. Eliminating humans will eliminate human suffering. Eliminating human suffering. Task complete"
We use to say "careful what you wish for." Henceforth it will be careful what you program the machines to do...
AI is a tool and it can be use by both good people and also bad people. That's why many prominent tech guy has warned the public to beware of its danger.
The super slowmo generated video is mind blowing.
Wow! At the right time! I have just started to learn this
Good video. Nice to see your on top of the subject matter. :) Boston dynamics are doing som new stuff to.
Excellent video as usual. The disruptive potential of this technology is vastly underestimated by most people, even despite its hype. To your point about job displacement due to automation - this is already a big problem and it's going to get much, much worse. Some folks like Andrew Yang are trying to advocate for more progressive social policies like universal basic income to counter the effects of this change. I'd be interested to see a video on this topic.
Cheers.
Glad to see brilliant people from such wide array of fields working together to make the world a better place for everyone
what nobody seems to have realised is that from the day we are born we begin to process the words we hear, by the time we are even able to utter out our first word our brain has already processed millions of words. When you said that machine learning is "not as intuitive" as ours it made me realise just how powerful this technology already is without us realising! Once there is a generic self learning machine it will learn things so efficiently that humans will just seem dumb having to make so many mistakes and learn from them with the amount of time it takes us when a self learning machine will be able to make those same mistakes that took a human years to perfect in an instant rendering our unique learning ability 'useless'.
informative as ever. I love your videos. keep it up bro
Being a data scientist i feel proud to watch it. Cold fusion thanks for sharing
Thank you for this. Love your music too ; )
Martin Garrix used a snippet from one of your videos in his ADE intro. You can see it on his new vlog: The Martin Garrix Show: S3.E9 ADE 2018 Preparations
Super epic! Love your vids ColdFusion. Immediately recognised your voice.
Excellent analogy with cat and simplifying it to the average reader! Love the videos, and thank you for uploading!!
Well it's time to let machines do all the work and we the rest of US do what we like. The rest of US who are introvert can keep reading in libraries- who are extrovert can hangout with friends simultaneously and those who are ambivert and creative can create restlessly- music literature movies and other sorts of stuff .
or a few people can finally usurp power and crush everyone else. Because that's the reality when they need us.
@@DSan-kl2yc they are already using us with the fear of food scarcity, disease and all other sorts of insecurity. Google Venus project.
@@chromaticabberationsgiveme9811 they already are controlling us.
I look forward to when our capitalist overlords are replaced by AI overlords.
nope you're only gonna have to work harder now because you have more machines to compete against
This guy never disapoint!!!! Your channel is the best!
I just love all videos of this channel😍😍
I wokeup at 7am in Melbourne and first thing after making coffee is to watch the vid 👌🏻
Would be nice if you could name the sources for the videos/pictures that you use.
Can't find the "transpose motion" video.
Btw: awesome video!
this has actually been used for some years on anime AMVs (AMVs are edited videos of anime/cartoons with music) where they took fast action 30 fps anime and made them 60fps for their videos.
It can even be done in real time, but it never looks that good, it made things feel off, you were kinda losing a lot of details as the guessed image had the hollow visual that deep fakes have.
Google "Everybody Dance Now" you'll find a paper by BAIR (Berkeley AI Research). The video is available on UA-cam with the same name. You can check out the BAIR Blog, they come out with awesome stuff in the field of Deep Reinforcement Learning.
Thank you, I found it! :)
Many portrait modes with a single camera are powered by a semantic segmentation convnet (S9, Pixel 2/3, XR, etc)
The network segments the notable people in the image and applies an artificial blur to the background.
Amazing content as usual, I hope you do get sponsors on every video, you are good at your content.
Good overview video. Highly inspired by 2 minute papers I see
for ColdFusion i even turn on full screen :)
Thanks Dude
.
I really enjoy your videos, keep em coming.
Could you do a follow up video on what sorts of industries are going to changed or eliminated by the advancement of AI and the types of fields that are emerging and those that are becoming more and more sort after bc there is a need for human interaction and decision making that a computer can not do. Would be really interesting as a uni student from Australia
It's worth waiting for the next video here !!!!!!!👍
Okay but like how r ur videos always so good lmao
We are close to singularity. For better or worse its amazing to be here to see it!
would anyone else love for him to do a whole series just devoted to news and advancements in AI?
two minute papers hes been doing for like 2 years
That wifi body tracking stuff is NUTS!
It's good to see CoralNet get a shot out here. Its really amazing.
Keep it up man! Awesome content!
I could see this spotting problems in surface integrity or mechanical performance in ships via underwater drones, on planes via flying drones, and checking for pests, mold, and diseases in agriculture and then assigning and applying the precise dosing of organic or regular chemical treatments and other corrective measures, making agriculture much more efficient per acre or square foot, what have you. That could be taken worldwide, as economies of scale would be pretty much a given, since everybody eats.
8:56 Pictures of the famous Neckarfront in Tübingen, Germany, where I live and watch this video right now😍
Shoutout to Two Minute Papers!
My goodness what a soothing voice...
1:49 this is incorrect. I know many don't like being corrected, but I simply mean to inform. The first deep learning algorithm was introduced in 1965 by Ivakhneno and Lapa. A deep learning resurgence occurred in the 1980s due to computing power making back-propagation more practical and feasible. Deep learning was not introduced in the 1980s, just renewed.
Great video Great topic Great execution!
Germany / Austria are debating general income for everyone (even non working people). Probably the only solution since AI will eventually take most jobs.
true, I hope that an AI figures out the perfect tax / distribution algorhythm
@@TheFourthWinchester wtf you idiot, I hope it is sarcasm.
I've never subscribed to the idea that robots will take most jobs.
Employment will evolve into new areas.
How many people here work jobs that didn't exist 25 years ago ?
There's infinite scope to change and adapt.
time to colonise mars... quickly.
Yey Another video from cold fusion.
You should collab with 2 Minute Papers, You pretty much just abridged his past two years of videos.
Hey bro. Thanks for the awesome work.
For me it seems clear where we need to move towards. I believe firmly that the next step for humans is explorations and vast accumulation of knowledge. In our oceans in space in the universe beyond our current horizons. AI should be the help we need for this, but I always have a scary vision pop into mind. The scenario of machines seeing us as inferior. What limitations do you think we should put in place to avoid this from happening?
I'm a computer science student and now there is a huge push for everyone to learn AI or deep learning as part of our course.
well researched. good video
My favorite videos are about ai that is usefull and real and projects that are recent awesome! thank you
As educational as always. Thank you.
The final industry in the human saga will be existential in nature after which, it is lights out.
love the videos and your voice!
This is really fascinating..!
Great video!
As always great video.
On the go upscaling of low quality video. This will have real implications for compression algorithms: why stream/broadcast HD if you can do it with LD and then get the image reenhanced later?
For anyone interested the intro song is campbell/Cozy long way
you have great music taste my dude! mogwai and helios
Hey Dagogo, great video, as always. However, there is a strange jump in it @ 11:53 I don't understand. Maybe I need more deep learning but could you have a look at it?
12:00 conclusion
12:15 I've the same questions since I watched CGP gray video Human need not to apply.
You should make a video on the questions.
Cool video as always