Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Tesla Live Stream – Autonomy Day [video]

livestream.tesla.com

Post reply on HN
Print view View original

371–380 of 550 posts

Re: Tesla Live Stream – Autonomy Day [video]

#371

Post by grandmczeb »

cromwellian wrote: post #258
grandmczeb wrote: post #165

Earlier quoted context omitted.

Convolutions have a much higher arithmetic intensity than dot products.

But in terms of CNNs, it's basically dot product is it not?Can you give an example of a CNN where it isn't?

In a conv layer, weights are shared across many input features. E.g. assume a 1x1 conv layer with a 28x28x3 input. You only need to load 3 weights even though there are effectively 28x28=748 different dot products. In practice, the input and output activations can be stored on chip as well (except for the first layer) which means the ratio of operations to DRAM accesses can be incredibly high. For some real world examples, take a look at the classic Eyeriss paper[1] which finds a ratio of 345-285 for AlexNet and VGG-16 respectively. You can also check out the TPU paper[2] which places the ratio at >1000 for some unnamed CNNs. Compare that to your analysis which yields a ratio of 2.

[1] https://people.csail.mit.edu/emer/papers/2017.01.jssc.eyeris...

[2] https://arxiv.org/pdf/1704.04760.pdf

Re: Tesla Live Stream – Autonomy Day [video]

#372

Post by naikrovek »

threeseed wrote: post #282
SheinhardtWigCo wrote: post #259

Earlier quoted context omitted.

This argument misses the forest for the trees. Even level 5 autopilot will sometimes crash into objects that are clearly visible to most humans, as will lidar-based solutions. But if the robot fleet is an order of magnitude safer than the human fleet, then it makes rational sense to prefer a robot driver to a human.

Problem with Tesla is that it is crashing into objects no human would crash into. And doing so at high speed.

I dunno. I've seen perfectly sober, intelligent humans drive into some extremely stupid "situations".

Re: Tesla Live Stream – Autonomy Day [video]

#373

Post by DeonPenny »

navigatesol wrote: post #221
DeonPenny wrote: post #72

Earlier quoted context omitted.

But Navigant curriculum was very unscientific. There no actual quantitive reason Tesla is worse. Is was based mainly on business factors like go-to market strategy and vision.

As opposed to the "scientific", "quantitative" reasoning behind Tesla being the leaders in FSD?

Yes absolutely saying Tesla who gets camera data from it's a half million car doesn't give it an advantage is crazy. That not even including the fact it's the only company who can do its strategy. Google would need to get constant data and GM and legacy automakers would need sensor suites on all it's cars yesterday.

No one knows if Tesla strategy will work because they don't have the data collection in place.

Re: Tesla Live Stream – Autonomy Day [video]

#374

Post by DeonPenny »

navigatesol wrote: post #224
DeonPenny wrote: post #76

Earlier quoted context omitted.

Why would you say that? The car has LTE and connects to WIFI. It could easily send way more data than any care company at any time including over WIFI.

> It could easily send way more data than any care company at any time including over WIFI. Except that it isn't, and even Karpathy said the quantity doesn't matter, it's the data quality.

What they are sending way more data because from our knowledge GM and Ford are sending back 0 data and Waymo doesn't have half a million cars worth of data internally to pick from.

Re: Tesla Live Stream – Autonomy Day [video]

#375

Post by alyx »

sytelus wrote: post #336
sairahul82 wrote: post #114

Karpathy's presentation is really good. Watch it later if you get some time. The key points are - Telsa is using fleet for learning (Fleet learning). This has 3 components, a. Trigger infrasture that collects the kind of data telsa is looking for training. The rational here is we don't need massive amounts of same kind of data but needs right kind of data for training neural networks correctly. b. Data Engine which l…

Karpathy is great at presenting executive-101. However self-driving is much much more than image classification. You have a time component, you need to integrate information as it comes in to meaningful state (aka SLAM) and make sequential decisions that has potential to kill people. This is still unsolved research problem unlike punny ImageNet classification. Karpathy could have easily spent entire day talking about…

Good overview, I share your opinion.

Tesla is betting BIG on NNs.

In the presentation they say that we have NNs in our brain. Which is why we can look at one photo of a dog, and distill enough identifying information in order to spot the breed from then on. Training sample of 1.

Curiously though, they than say that NNs don't work like our brains do, and require A LOT of training data from all angles, etc. Training sample of N.

The presentation doesn't actually show, how they plan to train a single NN to encompass all the knowledge for self driving. In particular, the looooooong tail. But maybe they'll have independent models, say based on "observed" conditions, which swap in and out ... based on more NN logic.

But at the end of the day, there is no known (to me) examples, of NNs even approaching human ability. Tesla is bascially hoping we will believe, they will be the first to show this capability of NNs.

Let's wait and see, 2020 is not that far away.

Re: Tesla Live Stream – Autonomy Day [video]

#376

Post by naikrovek »

ajross wrote: post #347
threeseed wrote: post #285

Earlier quoted context omitted.

That people drive vision is so idiotic. Humans have far higher resolution sensors and the most advanced computer ever seen behind them. And even that fails at a far higher rate than we want self driving cars to be.

> Humans have far higher resolution sensors This is really not true. The phone in your pocket is just plain better. More total pixels per unit area, much greater dynamic range, able to sample over time periods an order of magnitude faster. The reason your vision seems better is because your brain is amazingly good at synthesizing a picture of the world around you. But all that data (the sphere around you is something…

I agree that, ultimately, vision and some IMUs will be all that's needed for an as-perfect-as-possible self-driving cars.

Your statement on phone cameras being better than the human eye isn't true today, though.

The human eye has FAR better dynamic range than any camera built to date, at any price point. This matters a great deal on sunny, cloudless days in a city with tall buildings casting dark shadows, for example.

Just like how the biological system of "stupid sensor, smart thinking" helps biological organisms, that's what is going to need to be done for computer vision as well.

About 1/3 of the human brain is dedicated to vision processing in some way. Think about that a second. One THIRD of the best, most powerful organic computer known is required for us to see what we see and we are still fooled by the things human eyes can be tricked by. It's going to take a lot of neural network training to duplicate that. Fortunately, the vision skills required to drive are a subset of overall vision capabilities.

Re: Tesla Live Stream – Autonomy Day [video]

#377

Post by AnthonyMouse »

michaelt wrote: post #279
SheinhardtWigCo wrote: post #259

Earlier quoted context omitted.

This argument misses the forest for the trees. Even level 5 autopilot will sometimes crash into objects that are clearly visible to most humans, as will lidar-based solutions. But if the robot fleet is an order of magnitude safer than the human fleet, then it makes rational sense to prefer a robot driver to a human.

I think you might have replied to the wrong post here? I didn't say anything about humans vs self-driving cars. I'm arguing about self-driving cars with LIDAR vs those without.

> I didn't say anything about humans vs self-driving cars. I'm arguing about self-driving cars with LIDAR vs those without.

The status quo is humans so the first question is whether it can do at least as well as humans even without LIDAR. It seems to be doing that, even if the things it does get wrong are different things than the things the humans get wrong.

The question of LIDAR vs. not is a separate issue which is purely cost/benefit. Even if LIDAR has some value, it remains to be seen whether that value is ultimately more than its costs, in the same way that we know how to make cars safer by making them weigh 8000 pounds and yet in general we don't do that because the cost is more than the benefit.

Re: Tesla Live Stream – Autonomy Day [video]

#378

Post by eanzenberg »

ajross wrote: post #347
threeseed wrote: post #285

Earlier quoted context omitted.

That people drive vision is so idiotic. Humans have far higher resolution sensors and the most advanced computer ever seen behind them. And even that fails at a far higher rate than we want self driving cars to be.

> Humans have far higher resolution sensors This is really not true. The phone in your pocket is just plain better. More total pixels per unit area, much greater dynamic range, able to sample over time periods an order of magnitude faster. The reason your vision seems better is because your brain is amazingly good at synthesizing a picture of the world around you. But all that data (the sphere around you is something…

Take a photo with any camera at night and compare with what you're seeing with your eyeballs.

Re: Tesla Live Stream – Autonomy Day [video]

#379

Post by dyarosla »

natch wrote: post #324
RivieraKid wrote: post #301

Earlier quoted context omitted.

Waymo operates the first autonomous taxi service in Phoenix with plans to expand geographically, no info about the others.

Limited routes, geofenced, not truly autonomous, mows down pedestrians even with a supervisory driver.

Source for the last two claims?

Re: Tesla Live Stream – Autonomy Day [video]

#380

Post by eanzenberg »

RivieraKid wrote: post #270

I watch Tesla fairly closely from both the bull and bear sides. In short, I don't believe Tesla is anywhere close to Waymo. They won't achieve FSD in 2020. It's important to see this event in the context of their significant demand and cash problems. Even enthusiastic Tesla investors like Galileo Russell suggested that Tesla is in a cash crunch and should raise money. Which hints at the main mystery about Tesla, why…

2020? Try 2040.
Post reply on HN