HN user
GDPR isn’t enforceable against foreign companies. It is enforceable against subsidiaries registered within the EU. Living in Germany means you are doing business with Google GmbH (or likely, the Irish subsidiary). Don’t want to comply with German law? Then Google GmbH must exit the German market.
Car sharing in Germany such as Bolt and Miles. The pickup and drop-off anywhere model is so much more convenient than the point-to-point or round trip model that zip car uses. They are cost competitive with public transit or private car ownership.
What sort of programming did you use to work ok your mobility? I’ve been looking to do the same.
It does indeed cover it, starting at the 35 minute mark. It’ll require some more digging if you want the primary sources they reference.
I’d agree with this assessment. The American liberal arts education can do a fantastic job at producing well rounded and well read students. At the very least exposing people to arts and humanities — or conversely to the sciences.
There has been a lot of work in this space, and it’s a lot of fun to read around and play with. The Magenta team at google developed a differentiable digital signal processing system that is what you are describing. Here it is doing some tone transfer: https://sites.research.google/tonetransfer/about
I thought immediately of LightOns work: https://lighton.ai/photonic-computing-for-ai/
And this is very much already a vetted approach. They use photonics circuits to implement random projections, which can be used to speed up t linear algebra in modern ML workflows. It’s no magic monocle, rather a pretty interesting co-processor racked away in some data center.
My understanding is that the Julia community is quite interested in having SIMD via e.g. AVX “just work”. I recall reading this post on it a while back: https://juliacomputing.com/blog/2017/09/27/auto-vectorizatio...
My take is that most modern PPLs have language bindings in JavaScript/Python/R because they are explicitly courting analysts/data-scientists/applied-statisticians, or they are taking advantage of modern technical stacks implementing auto-diff and co-processor routines.
Most pre- and post- processing (with the exception of visualization) should probably be a part of your model!
The article does reference that Chinese study, as well as a more recent one from researchers in Italy and Spain. It is great to see the scientific process at work — building evidence for a phenomenon. It also underscores one of the biggest benefits of preprints: faster dissemination of information.
The point still stands. Rental caps mean it is very difficult to find a place to live in desired neighborhoods. Of course, you can find places outside the ring, but those would have been cheaper anyway!
That’s not to say it isn’t sensible social policy. It is less disruptive to people who have lived there for a long time. But if you move for work, you will have to either pay a lot more than you would have in 2011, or get lucky.
Great timing: I just finished “Objectivity” the book she co-authored with Peter Gallison. It is a phenomenal work on the history of science, looking at the development of the concept of Objectivity throughout the enlightenment through the lens of scientific atlases. Her writing is beautiful, concrete and lucid. Highly recommended to any Hacker News readers.
This was a thorough yet concise overview of the Structural Causal Models framework that has been gaining in popularity recently.
Is there any chance the course material is available? I've been looking to better my Julia meta-programming. Most of the resources out there are geared towards other languages. It would be amazing to see some simple parser/interpreter implementations.
You have to look beyond CO2 emissions to assess climate change impact. The IPCC land use report from this past August does exactly this [1]. The associated reporting is also quite good [2].
The grandparent is right: cutting down forest to grow cattle feed is a triple whammy.
1. https://www.ipcc.ch/report/srccl/
2. https://www.theatlantic.com/science/archive/2019/08/how-thin...
One thing this opens you up to is the average recommendation, which is biased towards inflammatory or click-bait content.
There was a log-normal version of the Galton board making the rounds last month:
https://stat.ethz.ch/~stahel/lognormal/bioscience.pdf
It really clarified where a log-normal distribution comes from: the consequence of switching a sum of random variables for a product.
One of the more remarkable results in 21st human genetics was that the top two PCs of a genotype matrix captures geography to an amazing degree [1]. This is the PCA of a N x M matrix where you measure M genetic markers in N people in a relatively homogenous population (Europeans in the paper above).
Fun fact: they are nearly the same implementation. See: http://markmail.org/message/ktzomp4uwrmnzao6
Those are fair points regarding energy expenditure for specific exercises. However, that doesn't take into account the effects of regular exercise on BMR. That normal exercise can elevate your BMR is certainly plausible, and many people have conducted studies that have reached this conclusion.
Metabolism is a complicated thing, and there are conflicting answers on the subject. Results differ under a variety of conditions (human, animal, forced/voluntary, etc.). The abstract in this review summarizes the state of the research nicely: http://www.ncbi.nlm.nih.gov/pubmed/14692598
Its great to see projects like these! First and foremost, this is the first project I've seen that defines a clear set of guidelines for structuring an analysis. So many of my projects have this "exploratory" angle to them that quickly grow into a unholy rats nests of directory trees. Each idea for another method to try, or another filter means that the se directories are filled with slightly different scripts and config files, with multiple copies of data sliced and diced different ways.
Lack of organization can stop any programming or coding project, but its clear that the authors understand what is specific to running an analysis. What particularly stands out is the automated pre-processing and caching of data at different points in a pipeline in a reusable way. Brilliant!
Can't wait to try this out on some real data. Hopefully the 'project' interface is as intuitive as it looks, and that the structure they impose is robust enough to handle a large project.
I'm no expert in compression, so its no surprise that I've never heard of xz. What are it's strengths compared to gzip, or b2z? Efficiency? Convention?
Nice simulation work! Initially I thought this was a one dimensional random walk problem where the answer is a function of n (as other comments have pointed out). In that case, there (n choose k) paths that take k steps in a single direction of n total steps, and each occurs with p(0.5^n). In this problem we are given n, but not k. Given a starting point, we can easily calculate the probability of crossing 0.
So my question is, how are you dealing with the initial condition?