144 karma · joined November 29, 2012
Next, let's assume that the micro controller powering the camera, transmission, wifi, and drone circuitry requires 200mA and 1.2V [1], or 240mW. Let's also assume this whole assembly is reduced to just a single motor, and draws better-than-ideal current [2], which would require another ~1.1W/377 minutes of flight time, or 175mW/hour, in addition to the 240mW/hour to power the camera and related circuitry. This is absurdly ideal, as current quadcopters require about 250 Watt to fly for 20 minutes with a weight of 600 grams, or 750 Watt/hour, a two orders of magnitude difference.
Since we would have 150,000,000,000 such drones, we would need .175 Watt/hour*150,000,000,000 = 26.25 Gigawatt/hour to power this system (but in reality more like 112.5 Terawatt/hour). To put that in perspective, the Three Gorges Dam hydroelectric plant (the largest in the world) produces 22.6 Gigawatt/hour at peak capacity and 98.8 TWh/year.
Maybe solar can help? Recently flexible solar cells were fabricated with 12% efficiency and a power-per-weight ratio of 23 Watt/gram. To power our quadcopter, we would need about 32g of these new solar panels, and continuously sunny conditions (since adding power storage to a solar powered quadcopter would add weight).
Of course the bigger challenge is logistics. How do you produce 150 billion of anything? If it takes 1 second to produce 1 drone, that's over 4,700 years to get them all. Do you work in parallel? Do you make 3D printers to print other 3D printers to then print these drones? Even at a failure rate of 0.001%, that's 1.5 million drones failing at any given time. What about costs? $1/drone? That's the annual GDP of Belize.
tldr; Invest in solar cell research and 3D printing.
[1] http://arduino.stackexchange.com/questions/8659/measure-tota... [2] http://multicopter.forestblue.nl/lipo_need_calculator.html [3] http://www.nature.com/nmat/journal/v14/n10/full/nmat4388.htm...
This sounds like a security nightmare, that also significantly impacts local update times and makes for a huge repo.
His early audience was attracted to his deep research into geeky things and hacking his way into things, but that has been heavily watered down as he expanded into writing multiple books [1], a mailing list, a TV series, a podcast, ... the list goes on. Although given his growing audience I wouldn't blame him for this. Another criticism is that Tim is by far not a primary or even secondary source for any of the things talks about religiously. He often touts books that fail to cite sources themselves, which just snowballs into misinformation. And this isn't about highly opinionated issues, it's usually about well studied regiments like weight lifting, nootropics, sleep, and memory. Why keep relying on Tim if I can go to bodybuilding.com, longecity, or bluelight and read cited reports firsthand?
[1] There was also added controversy to his character with speculation that Tim heavily relied on ghost writers, and even "hacked" his way to Best-Sellers lists [2]
[2] http://www.wsj.com/articles/SB100014241278873238643045783161...
They seem to have intentionally followed this design pattern, as hidden buttons often appear on hover next to adjacent buttons. I can't imagine how touchscreen laptop users feel using Asana. I won't even go into how their calendar view hides weekends, and how our team spent minutes finding the button to show weekends during a meeting.
"According to the equations physicists have settled on, gravitational waves would compress space in one direction and stretch it in another as they traveled outward."
LIGO is two sets of 2 L-shaped antennas spread far apart on the globe, so that we can compare the compression of space in orthogonal directions and measure the very short delay between the gravitational wave hitting the first detector followed by the second. In this case, that difference was 7 milliseconds, which is also consistent with the speed of gravitational waves (also the speed of light)
If the intervals spanned by x,y,z do not overlap, then your statement does not hold unless local minima are evenly distributed over the entire region, which defeats the purpose of searching for minima in the first place. If the intervals spanned by x,y,z are identical, then gradient descent can be reduced to just searching over f(x) by symmetry. You are right that there are m^n minima, but they are found in (nm)^1 steps. If the intervals spanned by x,y,z partially overlap, you can further reduce the search space to just f(x) and the union of the intervals.
I don't see how your example shows that gradient descent is likely to fail for any classifier with a sufficiently high number of independent parameters. Am I missing something?
Web developer - FREELANCE and REMOTE
Recently funded startup looking for developer to help relieve pressure from our main dev. The team is living as digital nomads throughout the US, so you can be remote or join us on the journey. Preference for freelancers on a project/hourly basis. Communication is key. You will not be the only developer on the team, so our expectations and tasks will be reasonable and informed.
Our stack: * is a small webapp running on AWS services (EC2+RDS+S3) that consists of nginx + postgresql + php + cakephp + javascript (jquery and angularjs).
We are looking for a developer to:
* help in finishing the implementation of a completed UI/UX design using the above tech (already 85% complete).
* implement new features in the product pipeline
* bug fixes
Contact yosyp@coral.life
Recently funded startup looking for developer to help relieve pressure from our main dev. The team is living as digital nomads throughout the US, so you can be remote or join us on the journey. Preference for freelancers on a project/hourly basis. Communication is key. You will not be the only developer on the team, so our expectations and tasks will be reasonable and informed.
Our stack:
* is a small webapp running on AWS services (EC2+RDS+S3) that consists of nginx + postgresql + php + cakephp + javascript (jquery and angularjs).
We are looking for a developer to:
* help in finishing the implementation of a completed UI/UX design using the above tech (already 85% complete).
* implement new features in the product pipeline
* bug fixes
Contact yosyp@coral.life
Does gradient descent apply to problems with many local extrema, or is it restricted to purely globally convex problems?
Looking for a competent remote iOS developer, with experience in Parse or other backend frameworks. XMPP familiarity is a plus. Web app experience preferred.
Willing to contract/freelance for lumpsum, or a share of equity, depending on level of interest.
I wish they had gone into detail about the numerical algorithm used to simulate the wave equation in three dimensions (Runga-Kutta order 4 probably?). As someone with a physics background, this is far from theory, or even original research.
It's interesting that they mention the necessity of Markov Chains for intersection calculations. Would anyone know if better techniques have been devised since the publication of this paper?
Take for example the Kinetis KL02 ( http://www.engadget.com/2014/02/25/freescale-kinetis-kl03/ ). With only 32KB of storage and 4KB of RAM, writing efficient code with a small footprint is important. These microprocessors can be used to make swarms of autonomous fly sized drones, and programmers that can cram more procedures into those 32KB are extremely rare and useful.