| Author |
Message |
mlohbihler
Joined: 24 Oct 2006
Posts: 2
10-24-06, 04:28 pm |
 Post subject: |
 |
|
I think it's irrelevant what language is initially used to develop the HTM. If the source is released once a usable system is created, other developers will port the code to the language of their choice. Then, once multiple ports are freely available, it will be become obvious what language - for whatever reason - is preferred by the community.
This all assumes, of course, that the source for the significant portions of the HTM system will be released when they are ready. Will this be the case? |
|
| Back to top |
|
| Author |
Message |
FreeSynapse
Joined: 11 Jun 2006
Posts: 39
10-24-06, 07:33 pm |
| Post subject: Probability of HTM being Open Source |
|
|
| mlohbihler wrote: | I think it's irrelevant what language is initially used to develop the HTM. If the source is released once a usable system is created, other developers will port the code to the language of their choice. Then, once multiple ports are freely available, it will be become obvious what language - for whatever reason - is preferred by the community.
This all assumes, of course, that the source for the significant portions of the HTM system will be released when they are ready. Will this be the case? |
I really doubt that will be the case, although it would of course be nice if it is. A developer porting the code to a preferred language would have to take into consideration that this is a product that will be undergoing routine updates by Numenta.
Maybe Numenta could consider adapting a tiered approach whereby there are two versions - one that's free and possibly open source, and a second that's not free and not open source, although I personally don't know about the viability of this. |
|
| Back to top |
|
| Author |
Message |
mlohbihler
Joined: 24 Oct 2006
Posts: 2
10-24-06, 07:44 pm |
| Post subject: Re: Probability of HTM being Open Source |
|
|
| FreeSynapse wrote: | | A developer porting the code to a preferred language would have to take into consideration that this is a product that will be undergoing routine updates by Numenta. |
A developer working within the original programming language would have to take the same consideration into account. Any updates he/she did would need to be factored into any new version released by Numenta (if Numenta did not accept the changes into the head of their source, which i would think is just as unlikely as with any other language).
In any case, Mr. Hawkins has repeatedly stated that Numenta's intention is to - to hopefully accurately paraphrase - release their development work to the community. I look forward to this happening.
If anyone knows of, or intends to start, a project to port the HTM to Java, let me know and i will pitch in. |
|
| Back to top |
|
| Author |
Message |
FreeSynapse
Joined: 11 Jun 2006
Posts: 39
01-14-07, 11:48 pm |
| Post subject: Additional Python libraries |
|
|
| In addition to the Python core, are there any specific Python libraries that will have to be installed or used, in order to use the platform? Thanks. |
|
| Back to top |
|
| Author |
Message |
Numenta-Phil
Joined: 28 Feb 2006
Posts: 62
Location: Menlo Park
01-15-07, 09:07 am |
| Post subject: Additional python libraries |
|
|
| Quote: | | In addition to the Python core, are there any specific Python libraries that will have to be installed or used, in order to use the platform? Thanks. |
We're doing our best to make sure that the answer to that question is "NO", however, if you want to analyze the results of your network, it would be best to have matplotlib installed as well. Other than that, we're relying on the built-in functions in the various Python builds out there. [/quote]
_________________
Phillip B. Shoemaker
Director, Developer Services
Numenta, Inc. |
|
| Back to top |
|
| Author |
Message |
Cerin
Joined: 26 Feb 2007
Posts: 4
02-26-07, 11:53 am |
| Post subject: Serialization/Persistence? |
|
|
| As I have experience with projects written in a mix of Python and C++, I look forward to your release. However, one of the problems I've had with C++ is its poor support for serializing/persisting objects. There are a few libraries, like Boost, that try to add serialization support to C++, but even then it's extremely difficult, especially when compared to Java or C#, which practically have this feature built-in. How do you plan on saving the results of your system? |
|
| Back to top |
|
| Author |
Message |
Numenta-Phil
Joined: 28 Feb 2006
Posts: 62
Location: Menlo Park
02-26-07, 11:57 am |
| Post subject: Saving system results? |
|
|
For the most part, we have two main items that people will want to save out: the network itself (which contains the description of the network, and, once trained, the set of spatial patterns, groups, etc.) and the results of the training or inference cycles.
For the former, we write everything out into an XML file. Basically, the network takes care of it for you.
For the latter, we have a simple effector (read: output node), that will write out a simple text file of the results. However, these effectors are easily written by any software engineer, so you can ultimately save the results in any way you see fit.
Stay tuned to the Numenta website for more information!
_________________
Phillip B. Shoemaker
Director, Developer Services
Numenta, Inc. |
|
| Back to top |
|
 |
Page 4 of 4 |
All times are GMT - 8 Hours
Goto page Previous 1, 2, 3, 4
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2002 phpBB Group
Please contact the board administrators if you have any questions regarding the OnIntelligence.org forums. |
| |
