Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Jeff Dean interview: Machine learning trends in 2020

venturebeat.com

Post reply on HN
Print view View original

41–50 of 60 posts

Re: Jeff Dean interview: Machine learning trends in 2020

#41

Post by ViralBShah »

Nashooo wrote: post #5

Does anyone have tips on how a European based developer with machine learning expertise can get involved with projects battling climate change like Jeff is talking about here?

I am happy to share my perspective on this. I feel a really good blog on this topic is Bret Victor's blog post - What can a technologist do about climate change?

http://worrydream.com/ClimateChange/

One of the clear recommendations is - contribute to Julia. A lot of what Bret said in 2015 has actually panned out. Julia has become a powerful language for scientific computing and machine learning. As a result it is being used in climate projects such as Climate Machine (MIT and Caltech). The Julia Lab at MIT participates in this project:

Climate Machine: https://clima.caltech.edu/ Github: https://github.com/climate-machine/ Julia Lab: http://julia.mit.edu/

Contributions to Julia packages (compiler, stdlibs, math packages, ML packages, parallel computing) will end up finding their way into climate research, because of the extensive reuse of code within the Julia ecosystem. Specifically capabilities such as Zygote.jl (https://github.com/FluxML/Zygote.jl) for differentiable programming have the potential to dramatically make it easy to apply ML techniques to scientific codebases. Compiler contributors are hard to come along, so all contributions to compiler technology are incredibly valuable. The DiffEqFlux.jl ecosystem in Julia is a good example of combining mechanistic models with ML (https://github.com/JuliaDiffEq/DiffEqFlux.jl). Hop on to the Julia slack channel or discourse to dig in deeper.

Another thing Bret Victor speaks about in his blog post is working with agencies such as ARPA-E on advanced projects. Julia Computing is participating in an ARPA-E project to bring these capabilities to many energy related simulation and development technologies. This press release gives a broad idea:

ARPA-E press release: https://www.energy.gov/articles/department-energy-announces-... Funded projects: https://arpa-e.energy.gov/sites/default/files/documents/file... Julia Computing press release: https://juliacomputing.com/communication/2019/12/09/arpa-e.h...

Re: Jeff Dean interview: Machine learning trends in 2020

#42

Post by mkolodny »

paganel wrote: post #10
narrator wrote: post #4

Somebody tell Jeff about Jevon's Paradox, or actually don't bother.

He probably does know about it because he seems he’s a well-read guy, but I think is too late for him and for people like him now: the pay they receive is too good for them to leave it all for some “principles” and on top of that I think the’ve also managed to acquire come cognitive dissonance traits that allow them to get out of bed in the morning and go to work without feeling guilty. Otherwise I cannot understand…

From the article:

> all the stuff we trained in our Google Data Center — the carbon footprint is zero. Because … basically, all of our energy usage comes from renewable sources.

Re: Jeff Dean interview: Machine learning trends in 2020

#45

Post by outside1234 »

mkolodny wrote: post #42
paganel wrote: post #10

Earlier quoted context omitted.

He probably does know about it because he seems he’s a well-read guy, but I think is too late for him and for people like him now: the pay they receive is too good for them to leave it all for some “principles” and on top of that I think the’ve also managed to acquire come cognitive dissonance traits that allow them to get out of bed in the morning and go to work without feeling guilty. Otherwise I cannot understand…

From the article: > all the stuff we trained in our Google Data Center — the carbon footprint is zero. Because … basically, all of our energy usage comes from renewable sources.

... which takes all of those renewable sources away from the rest of our energy consumption, so net net, their data centers cause more carbon.

Re: Jeff Dean interview: Machine learning trends in 2020

#47

Post by a_imho »

cookie_monsta wrote: post #38
GrryDucape wrote: post #21

I don't like the idea of a computer that can think for itself, I don't like the idea of computers will replace humans jobs, I don't like the way we are heading.

I wouldn't like the way it was heading either if Netflix could manage to recommend a movie that I might actually like. As is, I'm not that worried.

Hype does not care about whether it works as advertised on the tin. As long as it generates money you will be served buzzword whether you like it or not and you will pay for it. If lucky, the setback will only be financial.

Re: Jeff Dean interview: Machine learning trends in 2020

#49

Post by dwaltrip »

outside1234 wrote: post #45
mkolodny wrote: post #42

Earlier quoted context omitted.

From the article: > all the stuff we trained in our Google Data Center — the carbon footprint is zero. Because … basically, all of our energy usage comes from renewable sources.

... which takes all of those renewable sources away from the rest of our energy consumption, so net net, their data centers cause more carbon.

It's actually most likely the reverse. The more renewables are manufactured and used, they more they will be improved and made increasingly cost-effective, which will in turn make them more widely useful and competitive. It's a positive feedback loop.

This is what we have been seeing the past 10+ years, and of the main reasons why renewables are starting to be directly competitive with fossil fuels in some regions, even without subsidies. This is a huge turning point that we are on the cusp of.

Re: Jeff Dean interview: Machine learning trends in 2020

#50

Post by choppaface »

nl wrote: post #13
paganel wrote: post #10

Earlier quoted context omitted.

He probably does know about it because he seems he’s a well-read guy, but I think is too late for him and for people like him now: the pay they receive is too good for them to leave it all for some “principles” and on top of that I think the’ve also managed to acquire come cognitive dissonance traits that allow them to get out of bed in the morning and go to work without feeling guilty. Otherwise I cannot understand…

Jeff is right. Training a machine learning model for any common task is getting radically cheaper and more energy efficient. This is a combination of both better hardware (eg, Google TPUs) and better optimisation of training techniques (which is mostly done outside Google). Eg, DAWN Bench[1] benchmarks training of ReseNet50 to 93% accuracy. This used to take days. Now the FastAI group has shown you can do it in 18 mi…

Can’t wait for Cerebras. They can probably do an ImageNet in a couple seconds.
Post reply on HN