Hi Helge, I will be releasing the HTM.java Network API today which is very Stream capable and also uses RxJava (RxJava is a Java VM implementation of Reactive Extensions: a library for composing asynchronous and event-based programs by using observable sequences). You don't have to know anything about RXJava in order to use it; but if you want to do map-reduce patterned operations, it's available.
Network setup is concise and trivial, and does a myriad of Stream related things underneath the hood on the user's behalf - such as connecting and configuring stream input; encoders and classifiers etc. Once released I will be available for any and all questions so stay tuned! :-) @Helge, if you would like to get a "sneak peek" at usage pattern examples and discuss integration with your project, feel free to contact me. Cheers, David (cogmission) Sent from my iPhone > On Jun 5, 2015, at 4:23 AM, Helge Stahlmann <[email protected]> wrote: > > Started working on a NuPIC-Storm/Kafka integration (based on htm.java). > > Will post some updates shortly. > > From: nupic [mailto:[email protected]] On Behalf Of Chandan > Maruthi > Sent: Thursday, June 4, 2015 11:55 PM > To: Helge Stahlmann > Subject: Re: BigData and NuPIC > > Remember this ? VIKI :-) > > On Thu, Jun 4, 2015 at 1:24 PM, Pascal Weinberger <[email protected]> > wrote: > Welcome to the Future, my friends ;) > > On Thu, Jun 4, 2015 at 9:56 PM, Matthew Taylor <[email protected]> wrote: > "Big Data" is going to turn into "Unimaginably Massive Data" as the > Internet of Things comes online. Think about every sensor in the IoT > running an HTM model, predicting its own activity and doing its own > anomaly detection. Then imagine real hierarchies of these nodes so > that local groups of devices can be wired together to become a single > entity. Local groups become nodes in larger hierarchies, which monitor > their own health and make higher-level predictions. This is the > promise of HTM. > --------- > Matt Taylor > OS Community Flag-Bearer > Numenta > > > On Thu, Jun 4, 2015 at 12:25 PM, Román Martín González > <[email protected]> wrote: > > Thank you to everybody for the comments. The strategy I am doing is the same > > as for other technology that have developed over BidData Architecture with > > very good results: NeoSearch Technology (http://neosearchtechnology.com) I > > am going to use Spark to distribute processing different data streams, and > > most important, distribute execution the hierarchy of the HTM. > > > > > > -- > Regards > Chandan Maruthi >
