Start your deep learning journey with Andrew Ng here:
All the credit goes to the Lex Fridman Podcast. make sure to go and check these full episodes out and make it a habit to listen to this insightful podcast.
Best SQL Course Online (Medium article):
In this video, I’ve compiled some of the best machine learning advice from the Lex Fridman Podcast. I tried to feature the brightest minds of our time that are currently working on AI like Andrew Ng, George Hotz, Yann lecun, Ilya Sutskever, Sam Altman and Demis Hassabis amongst many others.
Hope you enjoy it and get enough motivation out of it to catapault you in your AI learning Journey.
00:00:00 Andrew Ng
00:00:33 Goerge Hotz
00:00:59 Andrew Ng
00:01:21 Andrej Karpathy
00:01:54 Andrew Ng
00:02:12 George Hotz
00:02:50 Joscha Bach
00:03:04 Ilya Sutskever
00:03:26 Yann LeCun
00:04:08 Demis Hassabis
00:05:00 Ilya Sutskever
00:05:40 Wojciech Zaremba
00:06:02 Sergey Levine
00:06:33 Yann LeCun
00:07:01 Ilya Sutskever
00:07:31 Yann LeCun
00:08:53 Sam Altman
Links to original videos:
Andrew Ng:
George Hotz:
Ishan Misra:
Andrew Karpathy:
Joscha Bach:
Ilya Sutskever:
Yann Lecun:
Demis Hassabis:
Wojciech Zaremba:
Sergey Levine:
Sam Altman:
YOUR THE ONLY PERSON ON YOUTUBE WITH THIS KIND OF VIDEO MAKE A MILLION OF THESE PLEASE!
Wow
i really don't have the patience for watching a full podcast video thx for this summary
A great video indeed! I was kind of feeling stuck on what to do next on my ML journey. This video really gave my some insights and motivation. Thanks!
Here's some advice: start now! Just do it. Stop wasting time trying to do things perfectly, because over time small and imperfect steps can really add up to a big improvement in your skills.
we got ML engineer motivation edits before gta 6
I am going to take the notes from this video in a Death Note 😂
I am learning ai
Business teams ask these kind of questions because they don’t know what ML can do. They don’t understand . It’s the job of PM who understands ML/AI to work with customers to understand what problems of the customer ML can solve.
why death note in machine learning algorithms ??😂😂😂😂
The last one just ruined the whole purpose of the video
Data is the king 👑
was expecting one from each of other turning awards receivers.. https://www.youtube.com/watch?v=HzilDIhWhrE&t=8s&ab_channel=AssociationforComputingMachinery%28ACM%29
A full vid of 11 min of advice, which sums up at last and says, Taking advice from others should be approched with great caution !! XD
Great vid btw 🙂
I watched this video and successfully applied ML to macaroni and cheese 🙏🙏🙏
Thank you for this 🙏 I really needed to hear the big hitters say my feelings on ML out loud
Is there any good site to read research papers from? I really want to start reading them. Let me know if u know.
"Listening to advice from other people should be approached with great caution".
That "Sam Altman" thing at the End of all the advices!
11 minutes of advices, and then the last advice is to not listen to advices, hhhh perfect timing.
can u provide me name of backgroung music please??
the logarithmic bars at the bottom are super annoying. what are you a cunt?
This video makes me want to sit on my desk and code Transformer from scratch
Teacher Andrew's advice about handwiting for recoding noted to our head seems great idea even it's like simple and dissapering tradition
Look who’s back dropping videos on YouTube! With consistency, you could totally take over the platform, man! ❤
the ghiath effect on sub not good bro let him simple
Death Note.
And the ending kind of negates the video, haha kidding.
The best advice is the last advice.
Arigato, Sensei.
00:04 Don't waste time collecting unnecessary data
01:27 Dive deep into problem-solving for effective learning
02:42 Becoming an expert through 10,000 hours of deliberate work
04:00 Choose creation over consumption for satisfaction and impact
05:26 Key advice for Next Generation
06:52 Reimplementing at different levels of abstraction is a powerful way to understand machine learning.
08:09 Machine learning embraces sloppiness through cost functions
09:34 Intelligence is inseparable from learning
Ending basically said we shouldn’t have watched this video at all haha
Thank you ❤
I like that the video ends with Sam Altman speaking on the need to be cautious with taking advice from people and knowing what one actually wants, after the long series of advices
Hey man, great video. How did you do the subtitles btw?
Amazing ❤❤
Cool
God! The last advice was the best for ME so far
i was a hacker, then an engineer, then a scientist. ML only works if you "do the science". test different things and KEEP RECORDS. Constantly test hypotheses and look at the rate of change. Everything else is just guess work. you can get lucky, or you can get good and consistent.
Deep learning is constrained by the absurdly inefficient backpropagation algorithm.
As an alternative try the Neural Network Builder, NNB, which is lighting fast with zero output error while controlling overfitting and underfitting at will. There is a free DEMO software with a graphical interface for Windows.
Since YouTube erases the links that I could place in this comment (it happened in the past) you will have to look for the NNB REFERENCES article in the HAL Open Science platform and other open research and academic sites.
The crucial fact is that classical neural networks (with any number of perceptron layers, and any number of perceptron units in each layer) are equivalent to geometric polyhedrons. Instead of using the data vectors to backpropagate the error of an initial network chosen in a blind manner by chance or "intuition", use the data to constructive and economically define a polyhedron K that fits the data, then trivially translate K into the three layer deep network N that solves your recognition problem. Yes, three layers suffice; forget the hundred layer nightmares. This approach also provides rule extraction, although this feature is not available in the DEMO.
Around backpropagation a large culture has grown that creates all sorts of barriers including intellectual inertia, personal success, financial interests, dedicated processors, silly investments like NVIDIA, and other flows of money.
It is a natural law that the backpropagation culture will resist renovation by looking away from the facts.
But facts will eventually prevail.
Best regards to all
Daniel Crespin
The handwritten piece of advice clip was wild lmao
video title: how to learn ml
video: death note
Don't take advice from others said by the same person trying to give advice on how to direct the future. Fool.
nice video bro <3
machine learning could help with mac and cheese production.
was that part on handwritten notes a death note clip?
this is such a cool video
gem of a mine of information
Nice compilation. Keep it up.
In a world of loss functions, your video is a series of local minima, each capturing moments of greatness inspired by unique experiences. These points of progress, shaped by data and lessons, guided by attention, feed the journey forward to the global minimum, propagating toward a true general representation in latent spaces.