astrophysician

14 ms·

61.
by astrophysician 6y ago
> I don't see the misunderstanding you mean. I said if you think a selection of the least-length or sparsest solution is ad hoc, then so is your choice of prior. Solving the same system without priors would be analogous (actually ma
62.
by astrophysician 6y ago
Yea right, it’s just so weird to deal with this problem as a scientist unless you do work that is regularly in the news and you have opportunities to learn how to better control what is written/said and deal with instances where the wr
63.
by astrophysician 6y ago
You forgot about the other half of the cycle: time passes, another wave of publicity bashes the initial hype, people assume the scientists themselves are responsible (even though they have been freaking out about their work being misreprese
64.
by astrophysician 6y ago
Nice yea I would agree with the vast majority of this, only thing I would add is that Adam/gradient methods are still useful in a graphical model e.g. to get a MAP estimate (and then you can get a rough posterior estimate using variati
65.
by astrophysician 6y ago
> I'd say pulling a "prior belief" out of the air for your problem, especially if it is conveniently chosen because it is one that is easy to work with, fits your rather broad definition of ad hoc too. This is the number o
66.
by astrophysician 6y ago
You’re right, but in ML/DS (and I would argue, even in other contexts you might not consider ML/DS) you usually are doing things in a Bayesian context (“what can I say about my model parameters given my observations and assumption
67.
by astrophysician 6y ago
A lot of tools in ML might be used in other places with other interpretations, and the intuition for L1 and L2 that you describe is not wrong at all, but (1) ML/DS is usually done in a statistical context so I would argue that it’s a g
68.
by astrophysician 6y ago
Nice! I took a look at your blog, looks like we have a pretty similar background -- I was an astrophysics PhD before my DS life; happy to connect!
69.
by astrophysician 6y ago
> By Bayesian methods I mean methods which also describe everything with statistical distributions including hyperparameters, and which solve for a distribution (or some hard-to-get feature of it like its mean or confidence intervals), i
70.
by astrophysician 6y ago
* understanding where loss functions and regularization terms come from allows you to reason about the right choice for your problem and possibly to extend/tweak them to suit your needs. Are you working directly with probabilities? The
71.
by astrophysician 6y ago
Oh so training time is virtually irrelevant to us and if it weren’t we would have to be a lot more careful about optimization methods and possibly which language to use. We also cannot use NN for the models we build (we are restricted to LR
72.
by astrophysician 6y ago
> Personally I'd advise against both SVM's and Bayesian methods for a beginner. Bayesian statistics is very much the deep end of the pool. I don’t know, I think it depends on what you mean by Bayesian. I would say understanding
73.
by astrophysician 6y ago
Adam, RMSProp, etc are just flavors of gradient descent so they’re useful on anything from ResNet to logistic regression. There are more flavors like natural gradient that are more useful for smaller problems since they require a Hessian ma
74.
by astrophysician 6y ago
Agree — people should do what they want and have fun learning. Its just a suggestion that’s also colored by my own experience. I will argue that if you’re going for a job in data science it is a bit more of a serious suggestion since you ne
75.
by astrophysician 6y ago
Exactly. I’m talking about how to orient yourself, PGMs are hard, sure but they don’t have to be outrageously hard. I would argue that if you do understand naive bayes (and it’s naivety) and you understand priors, then PGMs are just like ru
76.
by astrophysician 6y ago
I think PGM's are covered by a lot of "standard" ML texts -- someone else mentioned Murphy's book which is great and is humongous but is a good reference for pretty much every method under the sun. Bishop's Pattern
77.
by astrophysician 6y ago
Nice! There are many books that cover this, even the docs for Pyro/other libraries are useful, it just depends on your preference for how material is presented + your background. Bishop's "Pattern Recognition and Machine Lear
78.
by astrophysician 6y ago
If you're new to ML or datascience, I would recommend working to build a strong basis in Bayesian statistics. It will help you understand how all of the "canonical" ML methods relate to one another, and will give you a basis
79.
by astrophysician 6y ago
> If our goal is to flatten the curve to slightly below hospital capacity, current policy has flattened the curve way too much. Our goal is to simply avoid hospitals becoming inundated. How would it be possible to flatten the curve to &q
80.
by astrophysician 6y ago
6 month lockdown doesn’t seem necessary, but a lockdown until it’s possible to do contact tracing and containment does seem necessary, either way this is something that easily warrants a drastic response because the potential cost is so hig
81.
by astrophysician 6y ago
Again, you realize debating the true value of the fatality rate is a bit silly when we already know healthcare systems will be overwhelmed (and have been overwhelmed already!!) if we hadn’t and don’t continue to manage this with severe me
82.
by astrophysician 6y ago
Alright so lets unpack this argument, because (1) this is a pretty extreme oversimplification, and (2) the unstated suggestion you make is to quite literally let millions of people die world-wide who wouldn't have otherwise: > Data
83.
by astrophysician 6y ago
> Death is sad and terrible, but we don't shut down society because people die. Yea, but COVID has the potential to kill a lot of people, quickly -- are you suggesting it's a bad idea to "shut down society" to keep
84.
by astrophysician 7y ago
MATLAB is used by a lot of engineers, Mathematica is used by mathematicians/physics theorists, Python is used widely by a lot of different fields.
85.
by astrophysician 7y ago
High false positive rate (I’m talking like more than 1%) means (1) you are overwhelmingly unlikely to actually have the disease even if your test comes back positive but you don’t have obvious symptoms (2) the pool of people testing positiv
86.
by astrophysician 7y ago
Yea but not nearly enough of the population for this to meaningfully change the response, and now you’ve made the testing results completely meaningless so that we can’t track the disease. Maybe it’s not intuitive to people but we are actua
87.
by astrophysician 7y ago
Really? That test is meaningless. It screws up the numbers, and you might as well do a complete quarantine anyway, since you need to not just quarantine but isolate anyone who’s positive. Let’s say there’s 100,000 people in the US with co
88.
by astrophysician 7y ago
Yea agree about a demand surge when this is over, but I don’t think most businesses will be able to weather this well. Margins are usually pretty tight. A business already struggling will simply close rather than take out more debt amountin
89.
by astrophysician 7y ago
I mean, some people are overreacting about some aspects of this, but the main thing is: - the fatality rate for covid is high. People make a valid point about how we don’t know the “actual” (unbiased) fatality rate, that may be hard to
90.
by astrophysician 7y ago
Yea -- that's very interesting and insightful. It sounds like false positive rate is hard to nail down in practice, but I think from what you say there's a good chance it's well below 1%. I won't delete the comment for n

More ›