LOL. I agree I think the same. However we seem to be running with it as something really cute n all because he made big advances in the AI field right, so, we see that and it's funny lol.
My heart just feels right when seeing superintelligient people like Andrew and Ian talk in such down to earth way. We need more people with such intelligience and humbleness to change our world.
Currently reading Deep Learning. Iinitially, the math sections were a little intimidating. Now, Im much more comfortable reading 'math' and have a much deeper understanding of previous machine learning knowledge I already had. Thank you Ian and Yoshua!!!
@@bobsmithy3103 There's a github repository which sports the pdf copy of the book. You can find it by just searching "mit deep learning". I'm not sure if that breaks any laws though :)
I can bet 1000% Ian is an introvert that too deep introvert. His awkwardness is very clear. Man! they are contributing something in their life. Seeing them i feel whats the meaning of my life.
from this interview, these guys seem to have very admirable personas. I wouldn't know if the camera makes then extra vigilant. I admire their passion, to focus endlessly to resolve real-world problems. Why arn't political persons useful - Only loud mouth ego maniacs, with no self restraint want to be apart of politics.
00:05 Ian Goodfellow's journey into deep learning 02:02 Ian Goodfellow innovated GANs for generative modeling. 03:51 Goodfellow's determination to make his idea work paid off. 05:37 Games are at an important crossroads in deep learning 07:28 Importance of mastering basic math for deep learning 09:18 Evolution of AI and Deep Learning 11:19 Entering AI field without needing a PhD. 13:06 Importance of building security into machine learning algorithms from the start Crafted by Merlin AI.
The legendary interviewer realizing (at the beginning) and at the end of the interview that he was responsible for creating the GANgster! (or was it a hologram!)
GANs are great invention. I tried using CVAE but GANs give better results for high frequency components, but CVAE(conditional VAE) are easier to train, would had been great if GAN hyerpameter tuning would had been a bit easier,. just a wish list.
2 weeks ago when i saw "Ian Goodfellow" written in a article, i thought its misspelled and it is "I am Goodfellow" :p..................sorry for poor english
I wish I were smarter. I am about 100 pages into Ians Deep Learning Book, just up to section 5.7. Still a little tough for an old graduate like me. Apart from the maths, its a little frustrating that compute cycles for the majority if us is limited by GPUs, still taking hours to see if an typical CN algorithms are actually converging and performing. Deep Reinforcement Learning (e.g. DQN) still seems slow and fragile to iterate quickly to a robust solution.
You could use amazon web services and gain access to a fast gpu to test the CN algorithms. Check out course.fast.ai/lessons/aws.html to learn how to set it up. The cost is actually quite reasonable.
Math as we see it, is a way of imagining the universe. You either are there, or you are not. The whole challenge of Math education, is to get students to imagine the world through the lens of Math.
At 2:15, Andrew Ng mentions an old GPU paper that acknowledges Ian Goodfellow. But I cannot hear it. What old paper GPU paper acknowledges Ian Goodfellow?
Thanks for the interview, Andrew. Hey fellow learners, is anyone interested in starting a weekly ML research paper reading and discussion group with me?
@@AndresMilioto incorrect, he wasn't trying to take credit for himself. He was trying to take credit for his research group, who spend countless hours trying different experiments. The research is like a puzzle with thousands of moving parts. In the end, some random guy comes in and connects the final piece and takes all the credit. How would that feel? Just imagine, you worked hard on some project but your manager takes all the credit, how would you feel then? Jurgen is a genius! He deserved the Turing. I have a mad respect for German scientists. American scientists not so much.
10:48 INDEED!!! Actually the societal, security, and policy aspect is one of the most fascinating aspects to me because this is often what gets most neglected when it could easily be argued that these topics are by far the most important. Let's not unwittingly create another "atomic bomb" without considering all the implications on society and human life just because we can. :'(