Profile cover photo
Profile photo
Nikolaj Berntsen
255 followers -
Out-of-the box thinking IT-Professional with a Ph.d. in physics. Co-founder and CTO of GenieBelt.com
Out-of-the box thinking IT-Professional with a Ph.d. in physics. Co-founder and CTO of GenieBelt.com

255 followers
About
Nikolaj's interests
View all
Nikolaj's posts

Post has shared content
Interesting idea, going to read. 
Building Safe A.I.: A Tutorial for Encrypted Deep Learning

TLDR: In this blogpost, we're going to train a neural network that is fully encrypted during training (trained on unencrypted data). The result will be a neural network with two beneficial properties. First, the neural network's intelligence is protected from those who might want to steal it, allowing valuable AIs to be trained in insecure environments without risking theft of their intelligence. Secondly, the network can only make encrypted predictions (which presumably have no impact on the outside world because the outside world cannot understand the predictions without a secret key). This creates a valuable power imbalance between a user and a superintelligence. If the AI is homomorphically encrypted, then from it's perspective, the entire outside world is also homomorphically encrypted. A human controls the secret key and has the option to either unlock the AI itself (releasing it on the world) or just individual predictions the AI makes (seems safer).

Post has shared content
Fun talk if you grok Google X and moonshots, but in particular there is much to learn on the NP hard problem of how to align liptalk company values with real company culture.
Here is a fun talk I gave on Building Diverse Dream Teams at Grace Hopper 2 months ago.

https://youtu.be/CRbvdpPZySo 

Post has shared content
And my Company, +GenieBelt (which makes for great company) wish all a happy X-mas too :)

Post has attachment
Merry X-mas all <X denotes the unknown variable>
Photo

Post has shared content
HAL lipreading, what's not to love :) Combine with Microsofts in lab achievement of transcribing human speach at same error level as expert human transcribers. Interesting user interfaces to come for the positive thinking of us :)

Post has shared content
Future looks bright. Love the point about simply making the tiles stronger can reduce cost due to savings on less broken tiles during transport. Smart on all basic dimension - still working from basic principles :)

Post has shared content
Bwah, dark energy seems here to stay if you trust knowledgable researchers analysis instead of reading headlined from microphone holding media - Thanks +Ethan Siegel for taking my cake away :) (This post since I earlier shared the non critical recitement of the article since I never liked dark energy).
“There actually is a nice result from this paper: it perhaps will cause a rethink of the standard likelihood analysis used by teams analyzing supernova data. It also shows just how incredible our data is: even with using none of our knowledge about the matter in the Universe or the flatness of space, we can still arrive at a better-than-3σ result supporting an accelerating Universe. But it also underscores something else that’s far more important. Even if all of the supernova data were thrown out and ignored, we have more than enough evidence at present to be extremely confident that the Universe is accelerating, and made of about 2/3 dark energy.”

Just a few days ago, a new paper was published in the journal Scientific Reports claiming that the evidence for acceleration from Type Ia supernovae was much flimsier than anyone gave it credit for. Rather than living up to the 5-sigma standard for scientific discovery, the authors claimed that there was only marginal, 3-sigma evidence for any sort of acceleration, despite having statistics that were ten times better than the original 1998 announcement. They claimed that an improved likelihood analysis combined with a rejection of all other priors explains why they obtained this result, and use it to cast doubt on not only the concordance model of cosmology, but on the awarding of the 2011 Nobel Prize for dark energy. Despite the sensational coverage this has gotten in the press, the team does quite a few things that are a tremendous disservice to the good science that has been done, and even a simplistic analysis clearly debunks their conclusions.

Dark energy and acceleration are real and here to stay. You owe it to yourself to find out why and how!

Post has shared content
Woohaa, just the tiles I have been waiting for. Look forward to price and availability. I'll nap a powerwall too but wait for model III^ (^ is transpose ;)). 
SolarCity and +Tesla just announced two new products that will transform the way people power their world with clean energy. If you missed the live webcast, watch it here: http://spr.ly/62638D2GW
Photo

Post has shared content
Now that's interesting! It means they would have to redo the exhibition at the national history and science museum in new york, ..., among other things :) Never really felt comfortable about this accellerated expansion that required 90% (ish, look it up) of the universe to be this mysterious new thing, dark energy, with no-one having a clue about what it was. Would love if we could forget that theory. Probably it would give us more orders of magnitude to roam the universe, which I think in principle is a good thing (tm) even though I most likely won't see it and probably would not want to stay around for that long.
No, the Universe is not expanding at an accelerated rate, say physicists
Back in 2011, three astronomers were awarded the Nobel Prize in Physics for their discovery that the Universe wasn’t just expanding - it was expanding at an accelerating rate.

The discovery led to the widespread acceptance of the idea that our Universe is dominated by a mysterious force called dark energy, and altered the standard model of cosmology forever. But now physicists say this discovery might have been false, and they have a much larger dataset to back them up. 

Earlier this year, NASA and ESA scientists found that the Universe could be expanding around 8% faster than originally thought.
By all accounts, the discovery was a solid one (Nobel Prize solid) but it posed a very difficult question - if the collective gravity from all the matter expelled into the Universe by the Big Bang has been slowing everything down, how can it be accelerating?

Since scientists first proposed dark energy, no one's gotten any closer to figuring out what it could actually be.
But now an international team of physicists from institutions say don't worry about it, because it probably doesn't even exist, and they've got a much bigger database of Type 1a supernovae to back them up.

By applying a different analytical model to the 740 Type Ia supernovae that have been identified so far, the team says they've been able to account for the subtle differences between them like never before.

They say the statistical techniques used by the original team were too simplistic, and were based on a model devised in the 1930s, which can't reliability be applied to the growing supernova dataset. They also mention that the cosmic microwave background isn't directly affected by dark matter, so only serves as an "indirect" type of evidence.

Instead of finding evidence to support the accelerated expansion of the Universe, Sarkar and his team say it looks like the Universe is expanding at a constant rate. If that's truly the case, it means we don't need dark energy to explain it.

Source & further reading:
http://www.sciencealert.com/no-the-universe-is-not-expanding-at-an-accelerated-rate-say-physicists?perpetual=yes&limitstart=1

Paper:
http://www.nature.com/articles/srep35596

#physics   #space   #research   #universe  
Animated Photo

Post has shared content
2018, ambitious, but the future is near, +Marianne Dissing 
Miroculus is building a 3D-printed device that will be able to use a small blood sample to diagnose early stage cancer. And they're open sourcing their code.
Wait while more posts are being loaded