krasin

57 ms·

661.
by krasin 10y ago
Lidar alone is not enough, as it relies on the certain reflective properties of objects around (and there's also solar light noise). But yes, they ought to add a lidar and more cameras so that there's 360 coverage with a good over
662.
by krasin 10y ago
https://www.reddit.com/r/videos/comments/4vupn6/a_turkish_tv...
663.
by krasin 10y ago
Well, that's the point: we can now (mostly, theoretically) fix the issues when they appear, instead of hoping they would never show up, and the costs of fixing will only go down over time.
664.
by krasin 10y ago
>This outlines one of the problems with genetic engineering. The debug/test cycle is a few lifetimes. Well, with CRISPR ([1]), we at least can hot patch. 1. https://en.wikipedia.org/wiki/CRISPR
665.
by krasin 10y ago
While I don't have any particular opinion on TOS violations here, I have to point out that it's definitely not stealing. At worst, it's an unauthorized copying.
666.
by krasin 10y ago
Interesting. I've made a quick search over Runet, it seems that basalt / basalt-plastic rebar is in large-scale production: http://nano-sk.ru/stekloplastikovaya-armatura/bazaltovaya/ They claim it's
667.
by krasin 10y ago
All / most of them are likely Servo contributors. Servo: https://servo.org/
668.
by krasin 10y ago
>Are you aware of any company in Oracle's position that did not sue Sun Microsystems, obviously
669.
by krasin 10y ago
There's https://www.google.com/work/chrome/devices/for-kiosks/ but I never tried it.
670.
by krasin 10y ago
These are the benchmarks supported by Soumith Chintala (Facebook AI Research): https://github.com/soumith/convnet-benchmarks They place TensorFlow performance on par with Torch (within 10%).
671.
by krasin 10y ago
Engadget ([1]) tells us: ...developed a display that can show up to 32,000 colors. [1] http://www.engadget.com/2016/05/24/e-ink-announces-a-color-b...
672.
by krasin 10y ago
Out-of-tree targets for LLVM are generally a bad idea given how fast LLVM is changed. Every change could break the out-of-tree target, and being out-of-tree, there's no way the upstream developers could notice that.
673.
by krasin 11y ago
>What does that kind of hardware for CV primitives have to do with deep learning? They have hardware convolutions on 12 SHAVE cores (kind of a DSP core). It means that the chip can run some useful subset of convolutional neural networks
674.
by krasin 11y ago
For comparison, Tesla Superchargers are 120 kW at peak (when charging an empty battery): http://insideevs.com/tesla-cranks-superchargers-up-past-120-... They are not wireless, though.
675.
by krasin 11y ago
NVIDIA Tesla P100 has 21 TeraFLOPS of FP16 performance by their words. So they got 8 chips there.
676.
by krasin 11y ago
>I think you mean it doesn't have a single reference to the paper. yes, thank you for the catch.
677.
by krasin 11y ago
The article doesn't have a single reference to a research paper. It's entirely possible to write a similar essay with an opposite title like "The research is clear, let's keep elementary homework", and it would be
678.
by krasin 11y ago
Contemporary phones (e.g. iPhone 6S) are capable of running GoogLeNet at 1 FPS, see, for example, this demo (mine): https://github.com/krasin/MetalDetector AlexNet will run at ~10 FPS, I guess.
679.
by krasin 11y ago
There's also https://dreamscopeapp.com/ which is on App Store for at least 3 months (may be more, I am not sure)
680.
by krasin 11y ago
Apple came to Russia relatively late (near iPhone 4, I think). People did buy lots of iPhones before that anyway. They would just buy in the USA and have it delivered via post (sometimes unofficial post). And Apple understands that. I don&#
681.
by krasin 11y ago
yeah, that was a bit strange. By choosing Little Endian, the author would achieve the ability to mmap an uncompressed image file to memory and then use it directly like an array in all contemporary hardware architectures. Not that it's
682.
by krasin 11y ago
Another supportive (for your point) evidence is the Tensorflow Android example: https://github.com/tensorflow/tensorflow/tree/master/tensorf... It runs the network locally on the device. It's fast e
683.
by krasin 11y ago
Mostly agree with your point. Word Lens for Google Glass was the only functional AR app, but a good one.
684.
by krasin 11y ago
According to the machinability ratings [1], Inconel 700 is rated at 0.09 (less is harder) and Inconel 702 at 0.11, while most of the Titanium alloys are above 0.15. There's a single Titanium alloy family (MST) in the table that is just
685.
by krasin 11y ago
SpaceX also prints its SuperDraco engines ([1]) using Inconel superalloy ([2]) powder, which is even harder to process with a mill than titanium. 1. https://en.wikipedia.org/wiki/SuperDraco#Manufacturing 2. https:&#x2
686.
by krasin 11y ago
Google participated with ReCeption / Inception v3 model, see the results: http://image-net.org/challenges/LSVRC/2015/results In classification task, MSRA and ReCeption show very similar performance: 0.03
687.
by krasin 11y ago
Nope. We need to make the copyright to last 5 years for exclusive monopoly, and more 10 years for making profit out of the copyrighted work. And make Mickey Mouse public domain.
688.
by krasin 11y ago
$ cat /var/history.log | grep 'add.*security.*future' | grep -v failure | wc -l 0
689.
by krasin 11y ago
What can be done with m̶a̶g̶i̶c̶ CRISPR, can be undone with CRISPR. I don't see anything really permanent here.
690.
by krasin 11y ago
According to Wikipedia ([1]), the world energy consumption is on the order of 20 TW. A typical Russian nuclear reactor outputs 1 GW ([2]). It gives us 20000 reactors to cover the current human needs. It seems that the math is correct. At th

More ›