You can actually calculate exact gradients for spiking neurons using the adjoint method: https://arxiv.org/abs/2009.08378 (I'm the second author). In my PhD thesis I show how this can be extended to larger problems and more complicated and biologically plausible neuron models. I agree with the gist of your post though: Retrofitting back propagation (or the adjoint method for that matter) is the wrong approach. One should rather use these methods to optimise biologically plausible learning rules. The group of Wolfgang Maass has done exciting work in that direction (e.g. https://arxiv.org/abs/1803.09574, https://www.frontiersin.org/articles/10.3389/fnins.2019.0048..., https://igi-web.tugraz.at/PDF/256.pdf).
Well to be more precise the script contains functions of course , each of which produces one of the figures and takes the path to the data / data as input. In any case the REPL has no place in such a workflow because I want to be able to check in the script at the end and have a reliable way of reproducing any of the figures I put into the paper later.
I and probably many others intend to develop reproducible figures. The way to do that in python / matlab is to have a script which loads data from disk and then produces a figure (a png / pdf). You then execute that file many times each time tweaking one aspect of the figure. Julia makes that workflow almost impossibly slow.
This seems very strange, why on earth would you pay that much to listen to music and who are you trying to impress? Is there some implicit assumption that you can meet otherwise unavailable women / men there? That seems extremely unlikely as well.
Yes I know and there are some people in permanent psychiatric care without any perspective for improvement with current known methods, some are kept almost like animals or permanently sedated.
There is plenty of evidence of inhumane treatment in psych facilities over decades, including human experimentation, drug trials, debilitating brain surgeries, electroshock therapy etc. Psychiatry is probably the single medical field that has caused more harm than it has cured.
There was a narrative that was pushed by Davos and others that advanced economies should be structured around services. I talked to economics students in Manchester at some point and they sneered at me, when I argued that Germany was in a much better place compared to most places in Europe because it hadn’t lost all of its industrial base. They called that thinking backward. One other reason I can think of is that a large majority of decision makers in Europe and US are non-technical and/or even openly hostile to practical aspects of industry. So everything operates with multiple levels of indirection between stakeholders and technical people.
One thing that helped me manage anxiety was to monitor my caffeine consumption more
closely. I found that if I went over a certain threshold (> 4-5 cups of espresso)
my overall functioning was pretty severely impacted. Sometimes this resulted in me
waking up super stressed and not relaxed at all in the morning and not getting
enough sleep overall. Now I have a strict no coffee/tea after 14:00-15:00 policy. If I feel particularly stressed after a work day I go for a run to really exhaust my remaining energy.
I also tend to make lists ~1-2 hours before going to bed, categorised by project,
for the next day, so that I can let go of all the tasks that I can't accomplish right now.
I really like Things (paid app for macOS / iOS) for that purpose, but also toyed
with org-mode and task-warrior.
I think the novel thing is that this is doing it dynamically, which bypasses the static analysis and makes it applicable to situations where static analysis wouldn't be useful.
I think it makes a ton of sense for things like EDA software, they have to run on beefy machines that are shared between users and the user experience over thinlinc / X11 forwarding is typically pretty atrocious. If you sell that solution to Cadence they would be super happy probably.
I think the diagrammer is worth leaving through a bit, even as a complete non-expert: http://cds.cern.ch/record/186259/files/CERN-73-09.pdf
These are the very practical rules of computing some of the most fundamental quantities in our universe, all explained in slightly more than 100 pages.
What I would prefer is a set of protocols that are interact in a well defined way with each others, like the Email or Usenet protocols, then you can write different clients for these and decentralisation is just a matter of running different servers clients that don't necessarily interact with each other. Something like the CapeProto RPC framework makes it relatively easy to declare the interface of a service.
My understanding of the political system in Sweden is that the administration is unusually independent and not very political. That leads to way more rational decisions compared to if politicians made the decisions. Also the infections were distributed very unevenly in the population with immigrants (Somali in particular) being disproportionally affected https://apnews.com/article/1d7916cf6e48b7a231b894ef9cda1a19.
That is a really absurd statement. If you can't vote, you can also not run for public office. The way you enact change in a representative democracy is to organise, create a party and then get elected, or to join an existing party and rise in the ranks. Some parties in Germany have a 50:50 male / female leadership split. I personally know several women that were active at the local level and ran for public office. A female acquaintance of my parents was state education secretary in the 90s for a while. None of that was possible in Switzerland on the federal level before ~1970.
You have also been number 1 in money laundering for African dictators, educated the current dictator of North Korea, never gave back the gold / assets that were stolen from murdered Jews, only granted voting rights to women in the 70s and have a history of fighting for whoever pays the most. Somehow none of these things have affected the Swiss Brand.
Yes it was meant as shorthand, also I'm not an expert, but I worry that applying evolutionary pressure on the virus in the way it is done right now might actually produce undesirable results.
Here is a good summary of what experts (which I am not) think: https://www.asbmb.org/asbmb-today/science/112920/will-the-co.... Regarding the smallpox virus (as per the article): It is a very durable virus, compared to a flu like virus, that means it will have a different reproductive strategy.
I really don’t understand why you would criticize the article for writing about elementary ideas like symmetry over and over again only to write your own introduction. Also your characterization of the Langlands program is completely off.
Getting a job at google is nowhere near as competitive as getting tenure, you have to live in an alternate reality to believe that. I know several people that went there after their PhD one of them got the lowest possible grade on their dissertation (rete). Both published no papers as first author.
On PC it is a pretty decent game, nice missions, very nice graphics and art. Personally I like it better than other open world games that I played. It feels more immersive than most of them, as long as you don't go out of your way to break the immersion. In GTA and Red Dead Redemption 2 it is far more fun to create havoc than to play any of the missions in Cyberpunk 2077 there is basically almost no point doing anything but the missions, but I still feel far more connected and invested in the character than in either of the two other games.
It crashed a few times on me but never in missions or while I was doing anything important. Have a 1080 GTX and 32gb RAM it runs very well at 1080p and pretty high graphics settings.
What is more some of the deep learning libraries on top of JAX (like flax, linen) are close to a one-to-one copy of PyTorch. However the underlying implementation is in some ways more interesting than the one of PyTorch and naturally accomodates higher derivatives and also the way that optimization and state initialization is implemented is more functional and "theoretically sound".
I tried to play GTA V, but found the story very off-putting and uninteresting. I couldn't relate to any of the characters and found some of them frankly disgusting. All the missions in Cyberpunk I've played so far are well made, some of them really good like the parade scene in Japan town.