usgroup

57 ms·

661.
by usgroup 6y ago
It reminds me of Descarte's argument that even if we were just brains in vats coerced to believe in reality by some evil demon's manipulation, we still "know" something because "cogito ergo sum". A simple count
662.
by usgroup 6y ago
I think there's a category error here. The PhD and the tinkerer are not doing the same thing when engaged in the same problem.
663.
by usgroup 6y ago
I think IQ or some variant of it - which is what leetcode sounds like to me - has an intuitive appeal because it offers a way of objectifying hiring and stack ranking people. All other things being equal, the higher ranked leetcoders are bo
664.
by usgroup 6y ago
In the case of a general product, spending some FB bucks and creating a landing page doesn’t necessarily validate an idea. Perhaps more importantly it doesn’t invalidate an idea if it doesn’t work. Consider focusing on just physical product
665.
by usgroup 6y ago
An antenna is not a non local interaction. You mean interaction via a medium between non local things ? Eg having eyes , ears, etc? I also don’t see the either or case making any sense here.
666.
by usgroup 6y ago
Compressive sensing
667.
by usgroup 6y ago
I don’t think there is an alternative to management by measurement if scalable is what you’re trying to be. That doesn’t mean you don’t have long term goals or nuanced objectives even if your measures are coarse, nor does it mean your measu
668.
by usgroup 6y ago
I don’t think this is correct . You can calibrate a threshold for a metric empirically without interpreting the metric probabilistically. Ie that z > N is empirically a good threshold for anomaly detection calls for no necessary further
669.
by usgroup 6y ago
As others have said, PhD has about as much worth as the papers you published during your PhD. If you want to join a research outfit at Microsoft for example, it’ll be a continuity of what you were doing as a PhD.
670.
by usgroup 6y ago
The next cut of Stan will likely use TFP as a backend. I think that PyMC4 will also. The Stan team wrote everything from scratch in C++ including their own autodiff code which many regard as quite a stretch in terms of long term maintenance
671.
by usgroup 6y ago
Stan will get your from a prior to a posterior distribution that is best supported by your data, but typically the posterior and prior distributions will not be of the same form, so there's no loop back to make. In the case that your m
672.
by usgroup 6y ago
Under the hood Stan attempts to find globally optimal parameter values for your function which you've expressed as a joint probability density. To do this it relies on the same MCMC theoretical results which indicate how the recursive
673.
by usgroup 6y ago
Bayesian modellers are typically scientists and Stan probably has more indirect than direct users. For example, rstanarm and BRMS are both R regression packages which use Stan which are wildly popular in Bayesian circles. They enable hierar
674.
by usgroup 6y ago
Stan is exceptional if what you need is a hierarchical Bayesian model, and if what you want is rigorous way of quantifying the uncertainty associated in the parameter selections in your model. Stan users are more often R users than Python u
675.
by usgroup 6y ago
I don't think that's accurate. Sampling based approaches scale badly with data so, although there are a few exceptions, if you're tackling the problem as a hierarchical Bayesian model - which is most often what Stan is used f
676.
by usgroup 6y ago
I’m actually not sure that’s what he meant. I think he meant that if you picked two planets to have life on them at random, and the pairwise distances between planets forms a uniform distribution, then the distance between the two planets r
677.
by usgroup 6y ago
Looks like it’s a recursive minimisation. I wonder what an optimal minimisation would produce.
678.
by usgroup 6y ago
Encore! You sir, deserve a prize for this one.
679.
by usgroup 6y ago
I think we are in the medieval ages v2.0. I think a larger historical arc will reveal that. So in 50 years nothing much will change; mostly better mouse traps!
680.
by usgroup 6y ago
The title got changed since I posted the comment (thank you!). “Scientists” doesn’t mean all scientists; I agree. But: “Scientists say the world is flat” “Two scientists high on crack say the world is flat” You’d agree that the latter provi
681.
by usgroup 6y ago
It’d be nice if HN could edit any and all titles that attribute beliefs to “scientists” as a collective. Wouldn’t you agree that it’s a bit sloppy?
682.
by usgroup 6y ago
“Learning Rust will probably not do much to solve that problem for you. It won't assist you in making the ontological leap from a tired stereotype into something sentient and real. You will remain a replaceable silhouette with no disce
683.
by usgroup 6y ago
Certainly nothing a bit of extra theology and geometry won’t fix.
684.
by usgroup 6y ago
Can you say more about what’s interesting here versus JAGS for example? JAGS is fully declarative, mature and useful. It has benefits (like being able to work out how best to sample automatically) but it also has huge amounts of expressiven
685.
by usgroup 6y ago
+1 but let’s see where your comment ends up ranked in the thread at days’ end.
686.
by usgroup 6y ago
Yeah thing is people now routinely have their first kid at 36. Just everything has shifted forward and you’ve no frame of reference for it because you’re living it. Your generation is going to do everything older.
687.
by usgroup 6y ago
From direct experience; learn optimisation by understanding a variety of optimisation problems. Don’t worry at all about how to solve them, just get very good at being able to express the problem as a parameterised function that can be maxi
688.
by usgroup 6y ago
Well backtests depend on everything including assumptions about your data, your trading reality, corrections for multi hypothesis testing, comparisons to a billion benchmarks, working out why your advantage works when it does, etc. I dare s
689.
by usgroup 6y ago
Are malicious packages a problem for Julia?
690.
by usgroup 6y ago
Thank you for the detailed answer. I guess I should just give it a proper try by converting an old project into Julia. So here is an example and my worries ... My project uses GDAL, PIL, SentinelSat libraries to collect Sentinel 1 imagery a

More ›