Видео

I feel the incentives now lead to throwing twenty names on every paper. Even if it was a simple straightforward idea done by one person, and his friend proofread it. This leads to forgetting all of the names, because you can't remember that many names. Also it's become out of fashion to name things after people. In previous eras we might have called it "Hinton's method" instead of "dropout". I try to just keep track of the last name of the first author. And use that name over the official one if possible.

Seems kinda of a FSM

Human inductive biases. All of them.

the real question is who is behind the VPNs. Is it actually Russia/Iran, are they just eating your data and selling it for more profit, are the ISPs behind this anyway essentially upselling nothing? hosting a VPN is nearly free - and you can do it for free via free tier aws and cloudfare anyway. It's all openVPN with a wrapper GUI what was also left out is the crazy sales, entry offers and deals they have. like 18 years free and 95% off for the first 6 months etc.

Current generation of ai can't solve it. Because it can't learn.

Hey man, thanks for the video :D I wanna ask you, do you know Jesus? He's offering eternal life to all who repent and believe on Him. I'm not here to force you to believe, but we will all stand before Him one day with an account to give of our lives. All who call upon His Name shall be saved❤

69th Like. Nice.

in the thumbnail you look like travis bickle from taxi driver 😂

This is great stuff! Excellent work.

Thanks for putting the word out, fellas. Doing god’s work

Could you pls read this paper TIMESNET: TEMPORAL 2D-VARIAT

love u r explanation...good job

Bullshit, the guys work hard under a lot of pressure and you post this saying that you don’t care about them, how stupid is this post? Beyond imagination

Oh, that makes sense. Thanks for this clip!

I imagine him slightly pushing down his sunglasses then looking them directly over the edge while saying slowly and with death serious voice tone " I need to be honest with you. That's is not f#$%"& possible"

love the pentagram

Ten most innovative foodtech Startups m.ruclips.net/video/YTsy1EQ_tZo/видео.html

Cant wait to see Yannic in a business suit, black tie and aviators.

That was both interesting and funky.

I endorse flying Yannic et al to interesting places.

This is really cool! Thanks for sharing! Just wondering if we can have future vlogs with more stable videos? This one shot by your phone is really shaky.

Try fewer moving shots, more stillish shots. The footage was a tad blurry. I mean, that's good if you want to imply that you were drunk the whole time.

Hey! I still use that cap too, reminds me of in person conferences =)

Damn! Is this why I couldn't meet you in person at the World Artificial Intelligence Cannes Festival?!?! I was actually pitching myself there too. Maybe next time? 😝

I guess it was a good exercise.

why all the porn music when exploring madrid? Is there something about the town that I don't know?

I have not thought about this too much, but I feel like generalized Procrustes could be used as a baseline for solving some problems, esp. where certain problems demonstrate patterns like symmetry since the solvers explicitly solve for that.

What about if one used the image to generate the weights of an MLP that can predict the values at each pixel of the predicted output shape? Seems like it could be an effective solution, but I'm interested in what you guys think

Im subscribin. You need P r o m o S M!!

There is an easier way to fill space with color by dragging it from the palette :) Thanks for such an amazing video. It was a pleasant surprise to see it here and in ML news.

On the repo last commit, they added the Sleep keyboard command: is with TAB

So u can play it on Mac? Pretty neat mini game actually

The only reason I remember author names: So I can quote the Paper in Latex without looking the Key up..

TAAAAAAB!!!

2/10, was told not to expect quality content but got some anyways.

danke dir. Freitag examen ueber neural networks

The implication is that custom hardware form google, facebook might privilage those institutions to certain lotteries though their techniques respresent an inferior path of progression. Often the computational cost of certain models respresents a bottleneck to their deployment. The difference in hard between research and production hardware might induce a technological hard freeze.

Can you make a short with GAN/Autoencoder image interpolation? Cat to Dog to mouse to cow. This video is too slow to be an effective short.

Just wait when neuromorphic computing becomes commercial

Transformers are kernels so this makes complete sense to me. However, reasoning is poorly defined. We tend to use it as a byword for like, the ability to do 0/1-shot learning. That may be fine, but I think we need to figure out what it is before we ask models to do it.

I bet there is soon going to be a paper, "Transformers are Hopfield Networks" because they do pattern search and retrieval.

I see no evidence that there’s a difference between this and what we call “reasoning”

If-Except-If trees coupled with neural nets might be a better option. Given the string 'orang' if g the o except if prior is n then e. You keep looking back for stored priors until there isn't one. Training is very greedy, try to predict the next letter, if it gets it wrong look at the letter prior to the last looked at, then add that exception to the tree. This can all be done very quickly using g+1ashtables (avoiding yt algor.) You get spelling correction and semi-readable sentences. I think text is very much a case for mixed symbolic neural approaches. Otherwise things get inefficient.

But how would you say reasoning is different from pattern matching? Do we have some sort of proof or hint that reasoning is something complex at all?

Define reasoning

Yeah but human may as well be just pattern matching in a broad sense. There is always the good old Chinese Room problem when trying to argue that AI is not "intelligent".

Interesting video format, is this TikTok or RUclips? 😅

Lmao the voice effects are on point 😭

Just curious. If author names are not the criteria, how do you tell whether a paper is worth reading or not?

Also wtf I cannot understand these comments 😂