It seems part of the problem is that only health insurance premiums that exceed 10% of your income are tax-deductible. Insurance through employers comes fully from pre-tax income. This amounts to a significant penalty to those obtaining insurance separate from employment.
Does anyone know why this is the case?
(Not an expert on tax law, so correct me if I’m wrong on this.)
A beautiful related result is the Crofton Formula [0], which says that you can measure the length of any curve by appropriately counting the number of straight lines that intersect it.
(My co-authors and I discovered a generalization of the formula that also holds in curved space [1].)
Interesting as the paper is, please link to the arxiv abstract rather than the pdf as that's more web friendly. Reader can always click on he pdf link there.
I'm especially intrigued by the "Timelike, Spacelike, and Lightlike" diagram on Page 10 of the linked paper where "lightlike" phenomena share an endpoint. Fascinating.
Author here— Yes, this is a great question! I think there's a lot more interesting work to be done here. It would be great if we could understand e.g. why layer-wise learning rates help with large-batch training of ImageNet (https://arxiv.org/abs/1708.03888), and maybe the per-component noise scale has something to do with it.
Another author here– I'll add to Jared's comment above that for long-running experiments (like the ones our Dota team runs), it can be useful to track this statistic in real time to see whether or not it would be useful to scale up the experiment.
Whats a cheap and unobtrusive way to estimate the BSimple version of the noise scale in real time? Piggy back on ADAM's moving mean and variance estimates? Edit: I see that Appendix A has a method for the multi-device training setting, but I'm thinking of single device training.
One of the authors here. Thanks for the comment! Yes, we mention this work and number of others in the blogpost and paper. This isn't the first (or the last) paper on the topic but I think we've clarified the large-batch training situation significantly by connecting gradient noise directly to the speed of training, and by measuring it systematically on a bunch of ML tasks and characterizing its behavior.
Somewhere here on HN I read that in russian there would be a special word for having connections with someone who knows when goods will be in stores. Maybe we need a word like that for every language.
Oh, you want Section 8? Well, fill out these 20 forms, be in the right office at the right date and time, and you'll be entered into a lottery IF the paperwork you filled out matches our unstated criteria and allotted apartments.
Oh, and if you don't get in this YEAR, there's a nice city park. And the cops only harass you every other night.
I often wonder how much damage the term "affordable housing" has done to the goal of true housing affordability.
Politicians can pretend to be working on improving housing prices, while in reality doing nothing of the sort. As far as I can tell, "affordable housing" requirements produce only false hope for lower-income residents and increased market rates.
I know where I'm from, we have an influx of foreign students, paying out-of-country education costs. I have no issue with people from wherever coming here to study. It's cool, since it brings in a lot of ethnic stuff, as we all benefit.
The downside is actually that, alongside city government allowing "luxury highrises" for the students. We're in the middle of Indiana, and these rooms are going for $1400/mo.
It's a combination of many different things; Indiana University owns a great deal of land and pay taxes on pretty much nothing, the city council refuses to say no to the luxury condo makers, the council refuses to allow low cost apartments or assist funding them, the state refuses to let bloomington to annex more land. And of course, NIMBYs are in many neighborhoods, but this is only one facet.
What's "affordable housing"? The common refrain is 'The Market will tell us'. Well, the market also says "if you can't find a good job, you're gonna live in the city park".. So I don't put much trust in that market religion. Takes people in decision making roles to solve this one - Government.
True, but there are some kinds of data that people are still uncomfortable sending to Google. Medical data is considered especially private, and aggregation of medical data is a huge obstacle to improving treatment using ML. I think this could be really huge in that space.
There is a movement to change consent forms (which patients sign as they enter a medical system) to permit larger sharing of medical data outside of its direct use. IE, you are offered an opt-in to permit your data to be analyzed beyond an individual visit, possibly for medical issues completely unrelated to any immediate medical problem. The consent forms are transparent, and opt-in- the health consumer is informed what their data will be used for, and they explicitly have to say it's OK (blanket consent, can be revoked).
I think this is a win, because the consumer has the choice, and if enough people do it ,the resulting aggregated datasets will have exceptional power to help solve global medical problems.
When I bring the screen closer to my face, the area where I can detect the movement becomes larger. Far away from my face, I can see only a few spinning. If this area corresponds to my fovea, wouldn't I expect the opposite? Perhaps larger spinners can be detected father away from the fovea?
Does anyone know why this is the case?
(Not an expert on tax law, so correct me if I’m wrong on this.)