Very cool. I tried using PNG compression, it does actually do better, slightly:
PNG : ~15.1 seconds, 83% accuracy
I also tried dropping in zstandard compression:
Zstd (level=3) : ~3.5 seconds, 88% accuracy
Much faster than gzip, at least. And iff I use (x1-x2)*2 instead of x1+x2 to calculate Cx1x2 that pushes zstd up to 93% accuracy.
I'm somewhat interested in the fact that if I stack the two arrays on top of each other instead of adding them together, I get absolutely garbage performance (<20%), however as far as I can tell that actually works well when classifying strings.
Re garbage performance on stacking: I had tried interleaving the values (creating a 56x28 img), it dropped one percentage point of accuracy when using gzip.
I think the Guardian article mixes this up slightly. In the linked journal article they describe how when she initially presented with her symptoms she was treated with prednisolone, causing the immunosupression.
edit: and they note in the conclusion that the immunosuppression may be what allowed the migration to the brain
Possible but untested. You may be interested to see this recent study that draws a mechanistic link (broadly, microglial inflammation) between the symptoms of long covid and "chemobrain" following chemotherapy [0]. The authors do suggest treatments proposed to restore cognition post-chemotherapy may also be applicable to long covid, e.g. metformin [1].
Nir Barzilai and a team of researchers are leading a large clinical trial to test the benefits of metformin in slowing the aging process, which is upstream of so many diseases that yeah, it could just be a cheap cure-all. Sadly requires a prescription in the US. But berberine is the off-the-shelf equivalent.
Yeah agreed. Is berberine really an equivalent though if the mechanisms are different and the not knowing metformin’s complete mechanism although it has a pleitropic effect? Metformin is heavily studied in comparison too.
Do you also know a possibility to stop inertial/kinetic scrolling on two-finger-tap / rightclick? (left click works, but since I scroll with two fingers, it would feel much more natural on rightclick)
No worries. That one is a little harder. My understanding is that libinput doesn't handle kinetic scrolling at all[0], it is now implemented at the toolkit level. For Firefox, the "hold" gesture to stop kinetic scrolling is blocked on either the required event being backported to GTK3 or Firefox being ported to GTK4[1].
My only suggestion for the time being is to try scrolling back a tiny bit in the opposite direction rather than tapping. Works for me, lol.
So as a Wine end-user, it looks like this release means no more having to have multiple wineprefixes to support both 32/64 bit applications, and also no more need to ensure libraries like libpng zlib etc are installed, either 32 or 64 bits? That seems quite nice.
I haven't read this book, but this concept sounds very much like the notions of positive and negative liberty/freedom as described by Isiah Berlin [1].
In short, negative liberty is freedom from external interference, while positive liberty is having the resources and power to actually accomplish things.
"Such theoretical shifts set the stage, for Berlin, for the ideologies of the totalitarian movements of the twentieth century, both Communist and Fascist–Nazi, which claimed to liberate people by subjecting – and often sacrificing – them to larger groups or principles. To do this was the greatest of political evils; and to do it in the name of freedom, a political principle that Berlin, as a genuine liberal, especially cherished, struck him as a ‘strange […] reversal’ or ‘monstrous impersonation’ (2002b, 198, 180). Against this, Berlin championed, as ‘truer and more humane’, negative liberty and an empirical view of the self."
https://plato.stanford.edu/entries/berlin/
The idea of positive liberty is that you are free to become the best man you can be, but there is someone or a group who defines what 'best' is. And that inevitably ends in dictatorships or similar types of abuse.
I used to experience late day crashes and headaches quite frequently, and also chalked that up to excessive coffee intake. However, as someone who also spent little time in the kitchen (thinking I didn't have time for it), it turned out to be my haphazard style of intermittent fasting that was giving me the headaches. Now I spend no more than an hour or so each night making food with leftovers for lunch the following day, and my life has improved phenomenally. YMMV but I still drink plenty of coffee.
For anyone confused by the link - Cytoscape confusingly exists as both a javascript library and a Java desktop application. Cytoscape the desktop application is a very biology-focused network visualization/analysis platform, whereas Cytoscape.js is a pretty general purpose network/graph library. You can find Cytoscape.js at [https://js.cytoscape.org/].
Making their tracking data useless is kind of the point of the extension. Obviously it's well within Google's interests to stop people using it. I guess if you're privacy minded it's probably best not to use a browser made by a giant ad company.
I'm not sure how I'd feel if something like this became widespread and I was paying Google for clicks.
Well. I guess that's the point. But then surely it's fairly understandable for Google to try and stop them - at least on their own turf?
Google doesn't fund Chrome out of a sense of altruism - they want to have control over those aspects of the web-browsing ecosystem that potentially threaten their bottom line.
And here is that strategy behaving exactly as it should. You want the stuff Google has paid to build? You probably have to accept that "I want to automatically hide and click ads" falls under the banner of "maybe you should use a different browser then".
I'm somewhat interested in the fact that if I stack the two arrays on top of each other instead of adding them together, I get absolutely garbage performance (<20%), however as far as I can tell that actually works well when classifying strings.