nextos

7 ms·

61.
by nextos 6mo ago
https://sytse.com/cancer
62.
Going Founder Mode on Cancer(centuryofbio.com)
18 points by nextos 6mo ago|9 comments
63.
by nextos 6mo ago
> the reason it has been limited to those cases is drug development, today, is constrained by commercialization. That's a good observation, but I think it's an incomplete picture. Another important constraint is often regulator
64.
by nextos 6mo ago
You don't need an app if you don't want one. In a CLI, oath lets you calculate a TOTP. But it's maybe a bit more insecure if you use the same machine.
65.
by nextos 6mo ago
The problem with mRNA vaccines for cancer is effectiveness. Vaccines already work well for prevention of relapses in e.g. tumors that have been surgically removed. They might also be great combined with early-stage detection via ctDNA. But
66.
by nextos 7mo ago
Exactly, see what is happening in Hungary. Controlling Hungary is enough to veto some support for Ukraine.
67.
by nextos 7mo ago
Yes, and the EU, due to this fragmentation, seems to be a fertile playground for all this unacceptable interference by foreign powers.
68.
by nextos 7mo ago
I think we really need to use sandboxes. Guix provides sandboxed environments by just flipping a switch. NixOS is in an ideal position to do the same, but for some reason they are regarded as "inconvenient". Personally, I am a hea
69.
Embracing Bayesian methods in clinical trials(jamanetwork.com)
115 points by nextos 7mo ago|18 comments
70.
by nextos 7mo ago
Exactly, that's the way forward. There are lots of old ideas from evolutionary search worth revisiting given that LLMs can make smarter proposals.
71.
by nextos 7mo ago
AFAIK, it's a bit more than hyper-parameter tuning as it can also make non-parametric (structural) changes. Non-parametric optimization is not a new idea. I guess the hype is partly because people hope it will be less brute force now.
72.
Science Has a Major Fraud Problem(thefp.com)
3 points by nextos 7mo ago|0 comments
73.
by nextos 7mo ago
I guess this depends on the problem at hand. But I was thinking about a typical hierarchical model with partial pooling and standard weakly informative priors.
74.
by nextos 7mo ago
These days, the advantage is that a generative model can be cleanly decoupled from inference. With probabilistic languages such as Stan, Turing or Pyro it is possible to encode a model and then perform maximum likelihood, variational Bayes,
75.
by nextos 7mo ago
> I’ve never personally worked on a problem that I felt wasn’t adequately approached with frequentist methods Multilevel models are one example of problem were Bayesian methods are hard to avoid as otherwise inference is unstable, partic
76.
by nextos 7mo ago
Find recruiters within your niche on LinkedIn. In the UK, lots of jobs tend to be filled in via recruiters, and they are quite helpful.
77.
by nextos 7mo ago
I recommend you contact your local data privacy office, as they usually take this quite seriously. Fill in a formal complaint and watch the slow but inevitable unfold.
78.
by nextos 7mo ago
It's interesting Lean is taking off in software engineering. Prior to the advent of LLMs & agents, Lean had almost zero use in software and was mostly focused on mathematics, with Isabelle and Rocq leading the way here. In fact, I
79.
by nextos 7mo ago
Not just TDD. Amazon, for instance, is heading towards something between TDD and lightweight formal methods. They are embracing property-based specifications and testing à la Haskell's QuickCheck: https://kiro.dev Then, alr
80.
by nextos 7mo ago
Is SAT/SMT and theorem provers solving yesterday's problems 1.12% better? Lots of the successes by LLMs that have been much celebrated rely on these.
81.
by nextos 7mo ago
Something remarkable and unsettling is how the age verification debate has popped up almost simultaneously in the US, UK, and EU. With the same logical fallacies. Pretty telling about how transnational lobbies and their interests work. Cont
82.
by nextos 7mo ago
What configuration on the ThinkPad?
83.
by nextos 7mo ago
In the US, cheap ThinkPads like E14 sometimes sell for a bit less when you factor in all typical discounts. They are good machines that run Linux well and can be repaired. In EU, and I imagine other markets, there's nothing remotely cl
84.
by nextos 7mo ago
For instance, under Yann's direction Meta FAIR produced the ESM protein sequence model, which is less hyped than AlphaFold, but has been incredibly influential. They achieved great performance without using multiple alignments as an in
85.
by nextos 7mo ago
CSP and Hoare logic were brilliant. He was a huge proponent of formal methods. He famously gave up on making formal methods mainstream, but I believe there will be a comeback quite soon. On generated code, verification is the bottleneck. He
86.
by nextos 7mo ago
It was edited again a few minutes ago and now displays Sunday, March 8th as his date of death.
87.
by nextos 7mo ago
There is very little information around, this is the most authoritative post I could find. There are some comments on X as well. According to this blogpost, he sadly passed away last Thursday, March 5th.
88.
Tony Hoare has died(lefenetrou.blogspot.com)
268 points by nextos 7mo ago|34 comments
89.
by nextos 7mo ago
I agree. The formalization is very impressive even if we consider it was done on a result that is already accepted as true and got a lot of help from humans to scaffold and build the structure.
90.
by nextos 7mo ago
Jeremy Avigad has discussed this in a short preprint [1]: Gauss’s success was built on almost two years of creative work, scaffolding, and planning by the human participants, and it would have been unfair to them, mostly early-career resea

More ›