therobot24

Love that students are thinking about how they can apply what they learn to real problems.

However, choosing such a charged topic needs to be managed by an advisor - @h_jackson_ is about to get a lot more than she probably bargained for in terms of people arguing good faith, twisting her results to their narrative, and several arm chair scientists that are going to tear her down despite probably never actually reading what she's worked on.

Is her analysis perfect, no. Is it likely flawed, yea probably. This is where a professor can try to tone down the flames and push toward what might be next in the science of getting a right answer.

Git link: https://github.com/lab-ml/nn

I think this is awesome and personally cloned the repo so i can browse through the docs later without adding 'yet another tab' to my browser.

contract award doesn't mean all 22B was spent, that often means the contract ceiling where money can be obligated to it as we buy

glad you switched to syncthing, don't know how anyone can recommend dropbox when syncthing exists

> This is irresponsible at best.

Not sure i agree, they make it very easy to export your info/passwords and are just returning to a previous business model. As another user here commented, it only took 15 min to switch to another option.

> but a widely-used password manager needs to provide more time for users to transition into another product if they choose not to convert to paid

curious how much time that would be

i find using slides or scribe notes from classes to be way more valuable -- professor webpages will often aggregate these

It's always frustrating when right-wing nonsense makes its way into HN

inference is so fundamental to ML there should be classes just taught from the perspective - currently you'll encounter it in a graphical models course and when discussing networks of course, but again, viewing ML from the lens of inference like pearl does should really be it's own required ML course

while book 1 looks great, it appears that book 2 is still very rough: https://probml.github.io/pml-book/book2.html

“What we found over and over was that something is happening with the mitochondria regulation that throws everything out of whack.”

if a ml algorithm could influence me to work out more, i would accept the consistent nonsense ads about weight watchers

It'll probably come back in some form. Totally agree that too much emphasis was on the test score and that it didn't actually represent the student and their likelihood for success at the university. This is just Goodhart's law in effect, "When a measure becomes a target, it ceases to be a good measure."

However, the reason it'll be back in some form later is because you need quantitative measures to help sort the stack of applications otherwise the overhead becomes too much to deal with.

I'm torn on if i'd even want to setup this type of anywhere connection to my vehicles. On one hand, remote configuration, updates (maybe just use uptane?), and debugging is attractive, but the other hand is now you have a new threat vector that can cause a lot of harm if exploited.

is this an advertisement?

hug of death for me, will bookmark and see if it's up later

i think zoom is also the problem

I'm really surprised that binary CNNs aren't more popular - i understand that you're taking a decent hit in performance for CPU usage & a GPU really isn't all that expensive - but to me it's really interesting. When XNOR came on the scene in 2016 i thought they were really going to shake up deep learning and we'd see more binary CNNs pushed by Intel & AMD to counter the dominance of NVIDIA, but it never really happened.

>> Chromebook growth has been phenomenal, full stop. There is still plenty of headroom, because the price point continues to beat phones and laptop competitors. Note also the growing success of premium Chromebooks. Classic bottom-up innovation/disruption story.

I completely agree. Both my mother-in-law and cousins (that are college age) have switched to chromebooks because of the price point to accomplish what are basic needs for most laptop users.

>> [2] I won't ever buy another Google phone -- I bought a Pixel 3a and it completely broke by dropping it 3 feet with a screen protector and highly protective case, due to the very cheap display tech they used.

I've got a 3a and haven't experienced this. My kid has dropped my phone everywhere (from the cart at the grocery, from his high chair, his car seat, etc.) - note the phone isn't playing anything, he just likes to hold it and pretend he's talking to 'oma' or 'gigi' - and I just have a cheap rubber cover. To each their own though.

yelp doing something not shady or bullying? i'm very skeptical...

this is a weird collection of ML/computer vision/data processing topics

wavelets as it's own chapter, deep learning only has GANs as a single subsection, graphical models several chapters after the ML chapter...just weird arrangement choices

It's simple really, everyone has a right to privacy. The argument of having 'nothing to hide' is in bad faith.

your assumptions show more evidence of bias than there is of this judge being biased

phd in machine learning - it was worth it, but i had an awesome advisor and my own funding (via a fellowship) so my experience is not necessarily the norm. I would never recommend a phd to someone given the time commitment (almost lost my girlfriend, and now wife, and a lot of friends), but life is way better post-phd than if i hadn't done it. I understand the bind you're in. The phd will open more doors to more interesting work, your opinion will start to have more weight, and available interesting work finds a way to your desk. Just remember that you do pay for all of this with several years of hard, stressful labor - being an expert in a specific technology can get you to the same place without nearly as much of a personal time commitment.

looked at using airsim a few months ago to replace a product we'd been using for some time, problem was the actual vehicle dynamics simulation capabilities were subpar

>> I can see this being a fun interaction.

Oh for sure, unfortunately, it's the Facebook label that makes this cool product toxic. Since the Facebook brand is synonymous with poor privacy practices (Firefox even has a 'facebook container' to help reduce tracking), I don't see how someone didn't think to market this under a different parent company and make it 'facebook compatible'.

all of this sounds exhausting and doesn't highlight how little you would have to value your own time to do at the scale the author suggests

this has not been my experience with SBIRs

i feel like overleaf/sharelatex should be somewhere in this article

It suggests that you find low hanging fruit, don't do anything that will make waves, and only really think about the short term. That doesn't sound like you're operating in good faith.