1,852 karma · joined March 9, 2009
The churn rate will almost certainly go down on its own as your user base gets older, and consist more of long term customers. What you're usually looking for is to compute average customer lifetime, which is better done without assuming constant churn probability, see the linked post.
If BatchNorm, ConvNets, LSTMs, Dropout, ReLUs or any such essential neural net component came with such a clause, I could see the big players accepting the license, especially since it aligns with their publicly stated claim that they only file patents for defensive purposes.
One way to attack this would be if researchers outside of Google and Facebook started patenting their discoveries, but then immediately released all rights to it under a license that's valid for anyone who does not sue offensively. That way if Google wanted to sue for infringing use of invention A, they would themselves immediately be infringing on invention B and C. Quickly any large player would be entangled into a web where any offensive lawsuit is impossible. If we had standardized licenses with these clauses, we could stand a chance to stop this.
On some level, humans thrive under hardship. Social cohesion goes up, meaning and purpose comes easy. People who lose their job become less happy, and do not recover until they get a job again. Bill Gates wakes up and goes to his job, not because he needs money but because he needs to matter.
This worries me about AI and automation, universal basic income is needed, but doesn't solve people's need for having a purpose.
https://www.ovh.ie/dedicated_servers/gpu/ https://www.exoscale.ch/gpu/ http://www.cirrascale.com/cloud/plans.aspx http://www.nvidia.com/object/gpu-cloud-computing.html
Last link has a list of vendors. The one I used gave you a bare bones machine running some Linux distro, with no platform bullshit around it. Exactly what you need for deep learning work.
I am in the market for a cloud GPU offering, and I have to say the big cloud providers are very uncompetitive here, only offering these old, slow GPUs.
It's just not worth the effort at that point.
These days one should be careful when claiming things deep learning can't do. There are in fact systems that can imagine things it has never seen before.
Here's an example. It takes any text, and imagines an image for it: https://github.com/emansim/text2image
Here's a more recent example, with even higher quality images: http://arxiv.org/pdf/1605.05396v2.pdf
The traffic is encrypted, and no one has been able to MITM it yet, as far as I know. But Microsoft has confirmed in the press that yes, they do this.
(This usually evokes disbelief. If you think I'm wrong, please consider what you would do if it turned out I was right. Then, if you're on Windows 10, lookup your "speech, inking & typing" settings. This is where they ask permission to do this. It is turned on by default.)
I think the way we use technology today will be looked back on the same way we look back at naive cigarette smoking in the 1950s.
Modern app design isn't about creating things that are good for the user, but about creating want in the user. This is a problem.
For example, there are several studies showing that using Facebook in general makes people less happy. User happiness just happens to not be be necessary for Facebook to be a successful business.
Go to a developer conference by one of the big tech companies, and speakers generally aren't talking about doing good things for the user. They'll use euphemisms like "increasing engagement". There's concepts like "permission priming", psychological tricks to get the user to do what you want. There's books written about how to maximize app addictiveness. It's stuff that mildly screws over the user, and it guides the product designs that affect the lives of billions of people. It's not good.
Pretty remarkable that this line of reasoning was accepted by the brightest and best educated minds only 150 years ago.
I wonder if this is really true though. Some statistics:
https://www.fbi.gov/stats-services/publications/terrorism-20...
While Anders Breivik was shooting kids on the island in Norway, the Norwegian media were speculating that it was most likely perpetrated by muslim extremists. Even though we've historically had way more examples of terrorism perpetrated against immigration related targets by right-wing Norwegians (which was the case with Breivik as well).
The average Norwegian won't remember any of the mosques, immigrant reception centers and immigrant stores that were bombed in Norway in the past. I guess terrorism is quickly forgotten when it isn't your own group that is targeted.
> It differs from the established virtual DOM approach in that no intermediate tree is created (the existing tree is mutated in-place).
Direct DOM manipulation is something the DOM itself has had an API for since about 1995 (and there are plenty of nice wrappers for it). What value does this provide on top of that? Does it batch up changes for requestAnimationFrame()?
> So you're saying that the mt gox bank account was his own personal account? I wired money to his personal account??
> Yes
Maybe it wasn't correct in 2014, but apparently it was true when Barr worked there.