Hi Carsten,

Welcome! :-)

HTM.java is more recent, therefore it doesn't have the benefit of a wide
range of examples for its use. More examples are being added as time goes
by and community members contribute to the repository. The same is true for
the extent of documentation. Since HTM.java is a community sourced effort,
the documentation is basically left up to me and sporadic help by community
members which means that its taking a bit longer to get everything up to
speed.

There are 5 examples. QuickTest, HotGym, What does a fox eat?, and Breaking
News. 2 of which include working with the Cortical.io API but aren't any
different for that fact. Work with the HTM occurs in the same way
regardless of peripheral tools or APIs used - the thing to pay attention to
is the interaction with HTM.java's Network API (NAPI).

Functionally, HTM.java is exactly the same (as corroborated by extensive
tests). That is the "prime directive" of HTM.java - which is to be fully
compliant with the algorithms and results that the Python version uses and
gets.

What's missing so far is Swarming, and serialization - and there isn't yet
any HTMEngine style (multi-model runner). This functionality is on its way
as its only been really usable since June when the NAPI was added. There is
a roadmap for its development - but it isn't rigidly organized, the next
thing I will work on is serialization.

A community member has started a Scala implementation of a HTMEngine-type
application which uses HTM.java underneath, but that is a very recent
effort and I don't know the exact status thus far... That will offer the
ability to run multiple models in a scaled and clustered fashion - but
again I need to get to the serialization in order to make that as useful as
it can be. I won't be getting to that until after the HTM Challenge /
Hackathon is over because that is taking up my time for now. The project I
am working on (Nostromo) is a GeoTracking application - so there will be
another example HTM.java users can refer to.

If you have any questions, I am mostly always available by mailing list or
Gitter: https://gitter.im/numenta/htm.java

Cheers,
David

On Sat, Oct 17, 2015 at 7:26 AM, Carsten Schnober <
[email protected]> wrote:

> Speaking of HTM.java: are there any downsides in using HTM.java instead
> of the Python implementation? Missing features etc.? I couldn't find
> much information about that and the examples in HTM.java seem to focus
> on the Cortical NLP API.
> Carsten
>
>
> Am 16.10.2015 um 21:33 schrieb Matthew Taylor:
> > On Fri, Oct 16, 2015 at 12:14 PM, Carsten Schnober
> > <[email protected]> wrote:
> >>
> >> A way out of the Python world here might be to store the data after
> >> pre-processing in a simple text file
> >
> >
> > Or you can use HTM.Java!
> >
> > https://github.com/numenta/htm.java
> >
> > ---------
> > Matt Taylor
> > OS Community Flag-Bearer
> > Numenta
> >
>
> --
> Carsten Schnober
> Doctoral Researcher
> Ubiquitous Knowledge Processing (UKP) Lab
> FB 20 / Computer Science Department
> Technische Universität Darmstadt
> Hochschulstr. 10, D-64289 Darmstadt, Germany
> phone [+49] (0)6151 16-6227, fax -5455, room S2/02/B111
> [email protected]
> www.ukp.tu-darmstadt.de
>
> Web Research at TU Darmstadt (WeRC): www.werc.tu-darmstadt.de
> GRK 1994: Adaptive Preparation of Information from Heterogeneous Sources
> (AIPHES): www.aiphes.tu-darmstadt.de
> PhD program: Knowledge Discovery in Scientific Literature (KDSL)
> www.kdsl.tu-darmstadt.de
>
>


-- 
*With kind regards,*

David Ray
Java Solutions Architect

*Cortical.io <http://cortical.io/>*
Sponsor of:  HTM.java <https://github.com/numenta/htm.java>

[email protected]
http://cortical.io

Reply via email to