Recent Posts

Pages: [1] 2 3 ... 10
1
Pros
  • ConvNetSharp comes with a few working low scale examples
  • It has a GPU example for the MNIST data set
  • In comparison to other C# alternatives it has more features (e.g. convolutional layer, CUDA, like Encog does not feature that)
  • The author responds to issues/questions weekly and actually tries to help here and there

Cons
  • Lack of documentation
  • Lack of commented code
  • Struggling with exceptions concerning GPU and Volume usage
  • Tough to understand the code behind ConvNetSharp due to poor variable names
  • Single threaded CPU usage
2
Hi,
You did the pros & cons of implementing NN yourself, which I understand. But what are the pros of sticking with ConvNetSharp? (I mean, since it doesn't work nicely)
3
Progress is still sparse. I still didn't get a good result for the slot machine so I switched to the Apples&Poison Demo. Well, this revealed a huge lack in performance, because ConvNetSharp seems to be single threaded (no imports for using threads). The old port of ConvNetJS, which comes with the Apples&Poison Demo is running much faster, but is single threaded as well.

As ConvNetSharp features CUDA, I wanted to overcome this problem for now by letting the GPU do the job, but from that point on I'm pushed from one exception to another one. The first one was about setting the project to 64bit only and now I'm stuck on a CUDA exception concerning memory allocation.

ConvNetSharp does not come with any documentation or some comments in the code. What do you think? Should I stick with ConvNetSharp? The alternative would be to build an interface to Python or to implement neural networks myself. Of course the implementation takes much more effort, but has educational advantages. And I'd approach it using compute shaders in Unity that wouldn't limit the usage to nvidia GPUs.
5
General Software Talk / Re: Windows 10
« Last post by Mansde on Today at 07:51:07 am »
In order to solve this problem , I've failed many times
all I had to do was to add my microsoft account as a user, login once, and re-enter the Windows 10 . Windows 10 accepted the serial on the first try, and I have again a fully licensed Windows version.
6
General Chat / Re: The Kitty Hawk Flyer
« Last post by Freddy on August 21, 2017, 10:25:49 pm »
That looks the safest one so far  ;D

I did wonder about flying over water, but did not mention it. One for the lifeguards maybe ?

 8)
7
General Chat / The Kitty Hawk Flyer
« Last post by Art on August 21, 2017, 10:20:53 pm »
No, not the one from the Wright Brothers. This one is somewhat smaller and way cooler!


8
General Chat / Sort of a manned Quad-Copter...
« Last post by Art on August 21, 2017, 10:12:21 pm »
If you consider that those four circular units each house 19 motors and propellers!

I think I saw four blocks of 20 LiPo batteries (possibly more). While extremely smooth and maneuverable I think it only has about 10-15 minutes of flight time. Then there's the issue of safety. Sure it could go way higher but would you want to, without some sort of safety option?

Still, it's very cool when some people can create unusual and creative things.

9
I have a rather long distant plan to do logic induction (which is computationally intensive) distributed over nodes that use the same web operating system running inside browser (think of webassembly as a base). Maintainers (together with me) would compete for a percent of processor time and users would choose where to invest their idle processor time. An user could also pick up several projects and adjust some sliders to contribute some or all of them and be notified when a progress is being made. Maybe a processor time could even be a marketable item, but I don't believe one leacher could ever pay a mass of seeders, so I think the idea could work only for some projects of public interest.

How in the earth can that happen? Well, there is a thing I thought of. To start it up, users have to boot up the operating system into browser (something like Google Apps), which means there has to be enough web apps written for the operating system. Programmers could be motivated to write quality apps and libraries by collecting a share from special app store and forwarding a fixed percent (maybe 10%), distributed to all authors on whose work the app is based. Forwarding those 10% could be a guarantee for maintaining a quality programming libraries made for the OS, making the OS an attractive platform for programmers of all sorts. The circle then closes, and the money is flowing as a reward to inventive library makers which in turn attract end-app makers that actually earn everyone's bread. There is also a space for free apps, and 10% of $0 is $0. Once that there is enough users, interesting things from beginning of the story can be made, regarding to sharing processor time.

Just an idea, not much, but it could work.
10
AI Programming / Re: Who's afraid of C ?
« Last post by Zero on August 21, 2017, 09:30:44 pm »
 :tickedoff:  Literally every book about C is stuffed with bugged code samples. Unfkinbelievable.
Pages: [1] 2 3 ... 10

Users Online

22 Guests, 2 Users
Users active in past 15 minutes:
markwilson, Data
[Global Moderator]
[Roomba]

Most Online Today: 31. Most Online Ever: 208 (August 27, 2008, 09:36:30 am)

Articles