strebler

This problem is actually pretty obvious to fix and you don't even need a neural net - it can definitely be done "old school" computer vision.

Simply detect the presence of "flashing emergency lights" in the oncoming lane and disable autopilot when present. No object detector needed. The signal is so strong it's literally flashing extremely brightly in a regular, predictable pattern - any vision grad student should be able to figure this out.

Could there be false positives? Yep, but very few things will flash quite like that, and most half baked vision engineers can do this. The worst case is literally simply the driver taking over on occasion at night (maybe near strip clubs? lol)

I wholeheartedly agree. It's by definition an insult and I don't believe an academic measurement of stupidity (or intelligence) is a terribly fruitful endeavor. "My math says you're stupid!" is a place some people may go with this.

I try to think in terms of bias: "what biases do I possess that may cause me to yield an outcome that is negative?" or "what bias does person X have that lead them to those actions or words?"

I feel it promotes empathy, questioning and understanding. Not name calling. At least, it's helped me to figure out some surprising things about myself and others!

What, the elongated, shiny, rotating thing that was able to accelerate without any effect on its rate of rotation? Totally a comet, definitely, 100% guaranteed.

TF's deprecation velocity was way too high for my taste. Things we wrote would stop working randomly with their updates. I feel very similar to you about the models being "buried too deep" in their (ever-changing) machine. I much preferred how easy it was to hack Caffe V1 (once you got past the funky names, etc).

These days, I really like mxnet. Torch was a disaster, but Pytorch is much better. It's not bad in production, definitely my #2.

Interesting, however this just reflects an overfitted model. Fundamentally, this photo is not that different from a fashion influencer's posts.

We're a fashion engine and our system fully detected both of the people and all of their apparel (the person in question's shirt, his pants and it sees the "printout" as a low confidence handbag, as well as his shoes).

Not to say it would be impossible to trick our system, however, this method would not be sufficient given a good object hierarchy. Our system would have to have a triple miss across two methods - would need to miss his pants and his shirt and his body with the localizer, as well as his pants and shirt with the segmenter. And, if we were serious about detecting hiding people, you'd be surprised how gosh darn reliable the shoe detector portion is.

I don't see it being terribly feasible (and definitely not reliably so). Let's just say, it's not even close at all at this point. We miss zero of these things today.

I didn't quite understand, what is the problem with Instagram? It's too popular? Or just that it's not dropping in popularity vs Facebook (due to their recent adjustments)?

I wouldn't consider Instagram to be a messaging app - in fact they're spinning off another app from Instagram just to do messaging (because it's not exactly great at that).

It sounds to me more like Facebook might have some problem and Instagram is just doing it's own thing (as are Facebook's other services).

That's definitely possible they've already more or less decided. Either way, they're pulling off an impressive optimization to obtain maximum taxpayer funded subsidies.

That's incredible. Humalog is a quite similar low price in Canada. It's strange that it'd actually be economical for many American diabetics to just book a (yearly) flight to Canada and buy 12 months worth of insulin.

This is a really good article, I'm impressed. If you read only one deep learning article this month, make it this one.

I'm going to borrow that analogy of "teenagers perceptions of sex" - it's hilariously accurate for deep learning.

Exactly, do it or else he's going to take his ball and go home (or take his fund's money and go to China)

But don't tell him what happens if he ever tries to take his money OUT of China, we can let that be a "fun" learning experience for him.

I think they wanted to make it free but still deal with the fact that anything that isn't nailed down will get stolen in China. Even toilet paper - you should see the "rations" many hotels there give to guests for a night's stay.

The internet was actually fine for a long time (just proxy over SSH for everything). But they randomly "turn up the dials" on the firewall to block SSH and so you'll get intermittent outages. It's gotten progressively worse. We had a bunch of servers outside of China and I would need to basically round robin them to keep online.

I was actually somehow impressed by how sophisticated the firewall became over time from a pure technology point of view. Super annoying though.

I've lived and done a fair amount of business there. The car thing is true. I would say people there care a lot more about their immediate network of friends / family and a lot less about anyone else. It's very safe to live there, but at the same time everyone's trying to take advantage of everyone else - literal and metaphorical pickpockets are everywhere.

The inherent lack of trust makes workplace environments really difficult. Everyone's looking out for number one (themselves), to the point of damaging their own company for personal gain. You have no clue how hard it is to build a software dev team when ANY of the developers will happily copy the source code they have access to onto a thumbdrive and sell it to your top competitor for a modest payday. I highly respect anyone who builds a software team successfully in China, such a headache.

Frankly, the internet censorship was the deal breaker for me. No Google, gmail, facebook, twitter, news, etc, is just too much to bear.

But the food was really good, good weather and foreigners are treated like minor celebrities (everyone is really really friendly), so we don't experience real China like everyone else.

Hey, WoofCoin is going to change the world! Only WoofCoin uses blockchain technology to finally solve the age old question "who's a good boy?"

Based on the paper, I think there is sufficient evidence for "mice who have increases in sEH expression in Müller glia cells" to consider they stop taking Omega3. As I understand it, that's some sort of genetic condition?

This paper is not very clear for the layman and even if it was, it's very common to find completely contradicting results for similar studies in the biological sciences. Biology is not easy!

Anecdotally, there is a fair amount of papers related to Lutein being beneficial to T1DM and retinopathy but the best I could get out of my endocrinologist was "it doesn't look like it will do harm and could help, so go ahead".

The best I could say is to further investigate sEH in humans (which I will be doing also), speak with your doctor and make the best informed decision you can for yourself. Don't forget that you're not a mouse.

My primary questions will be 1) how does soluble Epoxide Hydrolase relate to T1DM? and 2) what is the likeliness that either T1DM patients in general (or me specifically) may have elevated sEM expression?

I think you're right. It was probably a good bootstrapping strategy at the start to get good designs into the system and also some marketing benefits from promoting top designs from contests. The more I think about it, contest-driven marketing is likely not sustainable - contests are more of a novelty and very hard to keep fresh.

Yes, having a store with selection is good for shopping. The contest is part merchandising and part marketing to show the "top hot new" SKU.

For any contest to work, you need good entries. The prize is the payout that motivates those entries.

I think a hybrid approach could likely work, where everyone's entry is buyable by default (and they get royalties). I'm not sure people would like if the winner(s) get royalties of other entries, but it might work.

One factor I'm realizing (which may have motivated the change) is that contests are a more of a "novelty" from a marketing perspective, so it may have sustainability issues. People easily get bored, especially if it's the same experience repeated over and over.

Why is it so surprising? From both the designer and shopper perspectives, the Old Way does seem better.

With the Old Way:

Designer: does N work to create a design and has X% chance of winning $2500. They focus on what they love (i.e. design) and can get a nice quick payout if they're good. This is even better than a lottery, because it has the same excitement, but talent is also a factor which is very motivating. The marketing is done for them. Competing head to head is also kind of fun.

Shopper: look at the top / winning shirts and buy one if they like it. Paradox of Choice is not a problem.

With the New Way:

Designer: needs to setup a shop, do N work to create a design, then M work to market the shirt, and needs to sell nearly 400 t-shirts to get to that payout. For the winners to achieve the same outcome, it's a lot more work. For most designers, I think it's safe to assume that the marketing part is nowhere near as fun as the design part. The excitement factor is totally gone.

Shopper: has to wade through tons of shirts to figure out which one they like. Paradox of Choice is signifcant.

Oh absolutely, I thought that type of active investor was so contrived that this was just an article written more or less for entertainment (to point out how hard it is to be an active investor).

Given the hyperbole and deity references, I thought the article was more for amusement than anything else.

I think he's just pointing out how hard it is to be a good active investor, basically exactly what Warren Buffet thinks (and won that bet over).

Does this mean I'm going to have to charge my clothing now too?

It's certainly possible that they stole the bitcoins and paid for a few days (a week?) worth of goods.

But 1) the article does not provide any real evidence to support for this theory, 2) it's not a sufficient amount to be a viable strategy, and 3) the evidence even linking the thefts to NK sounds somewhere between non-existent and very flimsy.

When the FireEye report's leading point states: "...there are no clear indications of North Korean involvement", it's not good support for such a broad statement as the title of the article.

Again, it's definitely possible, but too little information appears to be available to draw this conclusion.

I'm sorry, but a country of 25 million people is "dodging sanctions" with tens of millions of dollars in bitcoins?

That's not really an amount of money to dodge much of anything at all.

I completely agree. My company does image recognition in fashion and this dataset is both way too easy and not useful. I'd love to see a client with such an easy problem (telling between shoes and bags in inventory images).

We regularly get high 90's accuracy on real world images where the system has to auto-crop all on its own. Inventory images are far too easy.

What's more, the actual hard cases (like tiny shorts vs tiny skirt, or long blouse vs short dress) would be near impossible to do at such low resolution.

Tesla took the opposite approach of what I described (the Subaru approach).

Tesla took the approach that the machine is in control and the human has to detect and take over when the machine makes a mistake. That is known to be a very problematic strategy, it's easy for the human to get distracted.

What Subaru has done is to leave the human in control, the machine only steps in when it sees a problem.

It's totally different safety outcomes. In fact, not even the same problem - one is "self driving", the other is "crash prevention".

This has already started - look up the discount some insurers give to owners of Subarus that are equipped with EyeSight. The numbers show their collision avoidance is making roads safer.

Also, interesting fact: Subaru is achieving these impressive stats using stereo vision, not lidar.

You've hit the nail on the head - audio recognition is quite poor after (a lot of) concerted effort over decades. Driving a car safely is significantly more difficult (and dangerous).

That being said, Toyota's (and Subaru's) approach is the smart way forward - add sensors and capabilities that augment the human, but leave the human always responsible and in control. In order for a crash to happen, the human AND the machine BOTH have to miss it.

Once the machine is good enough (and we figure out what "good enough" means), then it can take over driving. But not before.

If the theory were true, then why were the two largest fires in recorded North American history in 1919 and 1950 (which MacDonald, a wildfire expert, missed)? 5 and 4 million acres respectively, significantly larger than 2011 or 2016 Alberta fires.

The data does not support the theory, the fires over the past 70 years have been smaller.

Sure, sure, let the fires burn. It's an easy argument to make when it's not your property. What about when the fire does over $3 billion dollars in direct damages, burns 1.5 million acres and forces the evacuation of over 80K people [1]? I'm fine with tax dollars going towards trying to prevent that.

[1] https://en.wikipedia.org/wiki/2016_Fort_McMurray_Wildfire