778 karma · joined January 23, 2011
Current: Staff Robotics Engineer at Chewy
Prior: Atlas @ Boston Dynamic; Robotics and Embedded Systems Engineering @ Amazon Robotics, Rethink Robotics, iRobot; Automation Engineering @ Ginkgo Bioworks.
Early Career: Electrical Engineer in Semiconductor Industry @ Applied Materials, Analog Devices, Allegro Microsystems
Similar looking model from same company: https://selinc.com/products/3355/
Their ingenious solution was to always run the system towards the warmer end of its spec, and so it included a program that would monitor the temperature inside the case, and would spawn/kill a bunch of threads doing compute intensive math in order to keep the temperature constant when the users workload wasn't enough!
I'm also not blind to the fact that there are still some deeply ingrained stereotypes, however subtle, about gender roles in our society, and I'm trying, perhaps poorly, to rephrase some points from the article (and others like it), in which the author very clearly names this: "the mental load bore almost exclusively by women."
This isn't a personal attack on you or your relationship. On the contrary, it sounds like you have a pretty healthy relationship, and I'm happy for you!
But the fact that this concept resonates with many people though, men and women, probably means there is something here that warrants more conversation for a lot of people, even if it isn't something that is a problem in your particular relationship.
This isn't about forcing a man "to fully adopt her perspective." This is about the idea that women are more often expected to be the ones to suggest a compromise, or to be the ones which have to point out these rifts in the relationship. More generally, to do the actual work of being in a relationship and keeping it healthy.
To flip around what you just said a bit: Why doesn't the husband notice that his wife is always picking up after him, or that she is not recieving acknowledgement of the work she has done, work that he feels should be acknowledged when he does it? Why must the woman be the one to notice there is a problem, to articulate it and to come up with a solution, as if the problem is her problem to solve and not a problem with the relationship, which consists of two people who are supposed to be partners?
EDIT: Link - https://en.wikipedia.org/wiki/Jacquard_loom#Importance_in_co...
(Because these kind of techniques aren't really enhancing the images in a way that gives you new and useful information: they are taking the low-res images as input, and giving you a plausible high-res image as output, based on it's training data. It is NOT however trying to say "this is the ACTUAL high res image that generated this low-res image"
Because it has 13 Co-sponsors, all republicans and no democrats. https://www.congress.gov/bill/115th-congress/house-bill/2359...
Meaning that if they lose their court case, they can still continue to behave badly, and not face any additional fines once they hit that measly total.
Oh, and just in case that $500,000 might hit some small "mom-and-pop" credit reporting agency, the total damage that can be awarded (again IN TOTAL TO THE ENTIRE CLASS), is the lesser of $500,000 or 1% of the net worth of the person/corporation the suit is being brought against.
Meaning we are basically saying "you can't actually bring a suit against them that could cause them enough damage to actually change their shitty behavior"
My gut says it's probably more realistic to do this for large scale maps (say between major cities, 100s of miles, where distance is pretty proportional to travel time). Once you get down to the city / street level, yeah it probably gets real dicey with high variation in travel speeds depending on the particular road and how routes are laid out, and you just aren't going to get pretty triangles fitting together nicely :-P.
In any case its a fun problem to think about!
0) ignore stuff like "infinitely quick" since that isn't realistic for what we are doing :-)
1) For the map I am displaying, pick a handful of points of interest to hse as controls, as well as my current location.
2) Use existing data we already have for stuff like direction routing to get travel times between all the pairs in our points of interest.
3) Draw a weighted graph, where each vertex is one of our points of interest and each edge is weighted by travel time. The initial position of each point should be it's geographic position, but we can then use an existing graph layout algorithm that uses an energy-based method (e.g. pretend the edges are springs) to deform the graph. This way each point will end up with its distance being proportional to travel time.
4) Using initial points as our input control points, and their location in the graph as output cobtrol points, use any number of image warping algorithms put there to calculate our warp / morph / deformation field.
5) Apply to initial map image.
I don't know how well it would work in practice but seems like a reasonable first stab at it! :-)
EDIT: Apologies for typo, fat fingers on mobile!
What I am saying is that the authors may have been a bit far-reaching in their conclusions, and that "trends in dating site photos" doesn't just mean "is there a rainbow flag present?", but also things like how photos are taken w.r.t angles, and lighting, and retouched, makeup, expressions, etc. (and yes VGG-Face is supposed to be resilient to some of these things, they also used Face++ for studies having to do with morphology, such as using facial contour features. But saying they are resilient to these things and quantifying that are two very different things).
To get to even a more fundamental question - Does the population of gay and straight individuals on a dating website necessarily reflect the gay and/or straight populations as a whole?
And once the authors start talking about the femininity or masculinity of faces, and whether certain features are gender atypical or not, they are entering very murky and ill-defined territory, which the authors even acknowledged when relating previous results on this front.
To be very clear about it: I am not saying that the study should be ignored completely, or that there were no interesting findings. Quite the contrary! What I am saying is that the conclusions being reached ("Our results provide strong support for the PHT, which argues that same gender sexual orientation stems from the underexposure of male fetuses and overexposure of female fetuses to prenatal androgen responsible for the sexual differentiation of faces, preferences, and behavior") are quite a bit of a leap from what was found in the research given the input data and methods, and I think should have been more accurately reported as: "interesting differences in social media photographs of gay vs straight individuals.", with a little less speculation by the authors on what might be causing these differences.
On peer review: 1) The uploaded version of the preprint that was linked is currently version 7. Version 1 did not contain the "this is a peer reviewed article" boiler-plate at the top. It's very possible that GLAAD/HRC saw the original manuscript without this notice. 2) Given that all 7 versions have been posted in the past 9 days, one might reasonably conclude that this is still a work-in-progress and peer review is still currently on-going, (the boilerplate being added in anticipation of the review being completed. Again, the paper is very clearly still being worked on.)
On misinformation: Yes I will give you that some of the criticisms might seem over-the-top to the HN crowd, but I think they are understandable given what GLAAD's / HRC's focus is, plus the fact that they aren't addressing "trained scientists." That being said, I don't think their criticisms are completely baseless. I think anyone with technical training, particularly in data analysis / machine learning, is familiar with the phrase: "Garbage In - Garbage Out." And I think it is absolutely correct to point out that maybe if your training data is dating site photos, predominantly of white people in a narrow age range, that you may not have exactly proven your claims that you have "advance[d] our understanding of the origins of sexual orientation and the limits of human perception." That maybe you've just learned some interesting features of current trends in dating site photos.
The tragic thing is that both sides (GLAAD & HRC vs. the researchers) claim to care about the privacy and safety of individuals. And kudos to them for thinking of the social implications of this kind of research! (We all know how good phrenology turned out) Because of these implications, I absolutely think it's right for us to question the conclusions of this research (that you can detect sexual orientation from facial morphology, and that this is evidence of the prenatal hormone theory of sexual orientation), given their actual methods (we threw a bunch of dating site photos at a neural-net classifier, and compared to self-reported sexual orientation)
All that said, I do think that the paper is likely to be more nuanced than the conclusions reported / the abstract make it out to be (have not finished reading the paper yet). But unfortunately, because of the way science is reported, where the headline is going to be some bombastic "AI can tell if you are gay or straight from just a picture!", there is probably some pressure to voice their criticisms just as loudly.
As much as I love printed materials for reference, unless it came that way, I'm not going to use a ream of my own printer paper for every reference manual or datasheet I have, when I could just have 1 monitor for coding and 1 with all my documentation readily available!
That said, I am really curious at actual lifetime costs of a system like this, and if it makes up for lack of efficiency by being simpler mechanically (leading to lighter weight, less maintenance costs, etc.) I think thats going to be the big thing that determines if this is a viable commuting option vs. rich person's toy.
And even if there is, the "non-evil" reason for keeping this as is might be that "enterprise customer doesn't want something that is easily disabled, and we can't justify cost of a separate design / fab run for a small minority of customers", vs. "Haha, now we have a backdoor into EVERYTHING!"
(Hell, even the proposals about disable via hardware switch, or having the ME on a separate component probably boil down to money. To an outsider, hardware companies are funny beasts in that they really start caring about things like counting pennies and minimizing BOM cost as much as possible. I've been in worried meetings where we were literally counting and debating pennies, and wondering how I got there, but it makes sense at a company level since those pennies add up when you are building thousands or millions of something! This may have boiled down to "motherboard manufacturer doesn't want to have to put another mechanical part on the board, and they are looking to buy a shitload of our parts, let's keep them happy")
Please note: I'm not saying, or even advocating, that this is a good thing, or results in ideal outcomes when I say "benign". I'm just using it in the sense that this likely wasn't a malicious decision, but probably one motivated by money at some point, by people who didn't necessarily see the harm in it.
(...With apologies to George Orwell)
What typically happens is that engineers do the best they can before release to model the process and experiment with building devices at different process "corners". This gives an idea of what the distribution of different electrical parameters will look like over the range of expected process variation (logic levels of FETs, sheet resistances, etc. etc.).
Obviously this data is imperfect, because there are a TON of steps in a modern semiconductor process, and a TON of variables in each one. (For example, I worked on Ion Implanters, and we recorded hundreds of variables for our customers during the run, from vacuum levels, to gas dosages to voltages and currents in the various accelerator stages).
As parent said, there is another group, who IMO are probably the original "big data" folks, usually called manufacturing or process engineers. They monitor the yields, binning and specific electrical measurements of all the wafers and parts coming off the line, along with process data for all the nanufacturing steps the wafer has gone through. These guys do the black magic of figuring out what needs to be tweaked in the process. At the higher levels it involves a mix of knowledge -- not only statistics and optimization, but also the semiconductor physics of how these parts actually operate, (since experiments can get expensive, you want to justify the knobs you are going to tweak!)
Before I got into robotics, I worked in the semiconductor world as a test and product engineer for a couple of IC manufacturers. A lot of my job involved figuring out how to test every parameter we were going to put on the datasheet, writing code to test it quickly, and then doing statistical analysis of the results over thousands upon thousands of devices to make sure they met requirements. It was an interesting experience. We used some very high end test equipment, made by companies like Teradyne, and for each part (or family of parts in some cases), we would make a custom board which would mate to the test equipment on one side, and had sockets for the part we were testing, along with any support circuitry required.
I think the problem with doing some sort of acceptance testing is that unless you buy the parts directly from the manufacturer or their distributors, it's very hard to verify that the part meets it's specs, unless you already have this kind of dedicated testing equipment and code, especially when you want to start measuring analog parameters (current consumption, logic levels, etc.), and don't think it would be worth the trouble in most cases.
(Military / Space may be a special case, where critical parts may stop being manufactured by the original manufacturer and they do have to do this sort of thing, but as far as I know, it isn't typical).
I come from an embedded background, where in some c++ projects, we would disable exceptions for various reasons, including the memory overhead, which is why I ask..