nonbel

106 ms·

931.
by nonbel 9y ago
>"Heavy and chronic cannabis usage has been linked to an increase in density, but a reduction in total volume, which resulted in a slight overall decrease in IQ." I wouldn't put much stock into these type of non-quantitati
932.
by nonbel 9y ago
Here is the activity survey: http://www.sdp.univ.fvg.it/sites/default/files/IPAQ_English_... The paper was confusing because they reported an average of 7 hours sitting each day and 1500 minutes per week of &
933.
by nonbel 9y ago
It looks like this was the study (emphasis added): Sedentary behavior associated with reduced medial temporal lobe thickness in middle-aged and older adults http://journals.plos.org/plosone/article?id=10.1371/jou
934.
by nonbel 9y ago
Where are you seeing that? Also, fine density/volume/mass, does it really change the argument? There is some optimal size for this region when it comes to certain functions, probably relative to the size of other regions. That siz
935.
by nonbel 9y ago
Is it bad to have a "thin" brain and good to have a "fat" brain now? Can't this brain region become too large? What if the area is enlarged due to inflammation/swelling and these people with "thinning&quot
936.
by nonbel 9y ago
There are multiple inference levels here. First, there is the statistical level, at which we are drawing some conclusion about the model parameter. This may work even for a misspecified model. Then there is the level at which you want to dr
937.
by nonbel 9y ago
I guess at its root the problem may just be how much compute is available to throw at the optimization. Alternatively there could be more efficient algos... I looked into but never fully tested this, it seemed promising: https://
938.
by nonbel 9y ago
When I tried this to choose xgboost hyperparameters it didn't seem to perform much better than random search while also adding another layer of hyper-hyper-parameters.
939.
by nonbel 9y ago
Why not just use regression trees, eg xgboost? The parameters aren't going to mean anything anyway.
940.
by nonbel 9y ago
This is impossible. If the model is incorrectly specified (does not include all and only the relevant parameters and interactions), it doesn't matter much what games are played with the math. Changing the model will change the estima
941.
by nonbel 9y ago
I've noticed it in other threads too so don't think it is just you. It just could be a coincidence...
942.
by nonbel 9y ago
Has anyone ever scraped HN and checked number of exclamation points per post and compared to story rank (eg time spent on front page)? On this page there are currently 6 exclamation points and 5 posts. This seems higher than usual, and actu
943.
by nonbel 9y ago
The wikipedia page didnt, and isnt expected to, answer my questions, which were about the methodology. For a given date given based on dendrochronology I wanted to know who collected the samples, how many were collected, which were included
944.
by nonbel 9y ago
This seems pretty dubious to me. In fact I'd think they would have less need for a good calendar since the proposed extremely regular harvests would happen regardless.
945.
by nonbel 9y ago
Sorry, don't remember anyone specific. I do remember there were only actually a few people/labs who seemed to be in control of the raw data though. Perhaps start here: http://web.utk.edu/~grissino/itrdb.htm
946.
by nonbel 9y ago
>"Quite a trick for some dude in a caravan 100 years ago to be able to simulate 85,000 years of radioactive decay." I don't think it is usually the constant decay rate assumption that can go wrong. It is instead the assump
947.
by nonbel 9y ago
Sorry, but here is what I was responding to: >"the consistency of Nile River flooding meant their calendars are likely the most accurate" You seem to be talking about something different now.
948.
by nonbel 9y ago
Sorry, but here is what I was responding to: >"the consistency of Nile River flooding meant their calendars are likely the most accurate" You seem to be talking about something different.
949.
by nonbel 9y ago
The way I remember is that 1 year -> 1 tree ring is more like an ideal than reality. You can have a series of cold summers with no growth or warm winters with continuous growth. Either can lead to many years -> 1 ring. Or you can have
950.
by nonbel 9y ago
I still don't get it. Why would that help you keep a calendar when you already have the seasons?
951.
by nonbel 9y ago
Interesting, but I don't really follow how flooding would be any more helpful than the seasons in general.
952.
by nonbel 9y ago
Another thing is I remember is a problem with people citing their sources. It was one of those topics with tons of BS making it hard to get to the actual data. Since you seem to know a lot about it, can you pick one artifact (I don't
953.
by nonbel 9y ago
I once tried to look into this and came away thinking it was all very circular (tree rings calibrated to archeology used to calibrate radioisotopes used to calibrate tree rings). There seemed to be a lot of room for cherry picking which dat
954.
by nonbel 9y ago
I'd think the main effect would be that the orbits of the remaining planets become unstable, haven't checked that though.
955.
by nonbel 9y ago
It is just the description of an idea I came up with without any implementation. I was leaving it purposefully vague, just do the "inverse" of what it says in that paper.
956.
by nonbel 9y ago
http://jmlr.org/papers/v15/srivastava14a.html
957.
by nonbel 9y ago
Here is an idea: inverse dropout. How would you implement that?
958.
by nonbel 9y ago
I don't mean this to be adversarial, but what exactly is it you do that would not be sped up by checking someone else's results directly before fiddling around and then trying out your own implementation?
959.
by nonbel 9y ago
You seem to be exceptionally well funded, and/or have few deadline constraints. Your strategy will only work until you get spammed with "good ideas".
960.
by nonbel 9y ago
What would be the effects of Venus crashing into the sun next?

More ›