Carin, Here is a wiki page with quite a bit of info about what others have attempted with NLP and NuPIC: https://github.com/numenta/nupic/wiki/Natural-Language-Processing
I'm going to point you to a few videos about HTM and NLP... Some of them are hackathon videos, some are presentations... (sorry about the bad audio) - NLP with NuPIC (https://www.youtube.com/watch?v=QNF-gONtSmA&start=1260) from 2013 fall hackathon sessions. This explains some experiments I was working on in preparation for this hackathon and tools I created for others to work with (unfortunately, most of these tools no longer work because Cortical updated to API v2.0) - Visualizing word SDRs ( https://www.youtube.com/watch?v=X4XjYXFRIAQ&start=3240) from 2013 fall hackathon demo. Where you can see the SDRs coming out of NuPIC's temporal pooling system that represents semantic ideas it has learned from word input - What Does the Fox Eat? ( https://www.youtube.com/watch?v=X4XjYXFRIAQ&start=7084) also from 2013 Fall hackathon. This is Subutai's demo the shows how HTM generalizes with word SDR input. - Famous (https://www.youtube.com/watch?v=O2VAP3B6t5M) is a 2014 hackathon demo from the Cortical.IO team that shows how HTM can be used with the Cortical.IO API to do help identify the context in which famous names are used. A couple of side notes... 1. Cortical.IO used to be called "CEPT", so you'll people talking about CEPT a lot, but they are really talking about Cortical.IO. 2. What Cortical now calls "fingerprints" we used to call "word SRDs". 3. The pycept API client is deprecated, and Cortical now has their own API clients in Java and Python --------- Matt Taylor OS Community Flag-Bearer Numenta On Tue, Oct 20, 2015 at 5:19 PM, Carin Meier <[email protected]> wrote: > I just finished up some explorations with clojure/ stanford CoreNLP/ Weka > classification of Speech Acts. > blog post - > http://gigasquidsoftware.com/blog/2015/10/20/speech-act-classification-for-text-with-clojure/ > repo - https://github.com/gigasquid/speech-acts-classifier. > > I was planning on trying out using HTM.java next to see the difference. > Are there any resources/ papers already that compare tradeoffs between > using HTM vs standard Machine Learning Classification like Naive Bayes, > J48, Random Forest? > > - Carin > > > On Sat, Oct 17, 2015 at 11:36 AM, Matthew Taylor <[email protected]> wrote: > >> This would be a worthy HTM Challenge submission! I don't have time to >> work on it, so... free idea! >> >> --------- >> Matt Taylor >> OS Community Flag-Bearer >> Numenta >> >> On Fri, Oct 16, 2015 at 1:27 PM, cogmission (David Ray) < >> [email protected]> wrote: >> >>> For anyone interested in HTM.java, I am available as a resource for >>> ideas or questions. I normally hang out in the Gitter chat room (a chat >>> available by web browser and is used for a lot of Numenta communities.). >>> Here's the link: https://gitter.im/numenta/htm.java there is a general >>> chat room for NuPIC which is here: https://gitter.im/numenta/public >>> which is very active. >>> >>> Please feel free to avail yourself of all the different resources for >>> learning about HTM technology. I also have a blog which I just started but >>> has a ton of good links which are helpful to beginners here: >>> http://cogmission.ai (see the menu) >>> >>> Cheers, >>> David >>> >>> On Fri, Oct 16, 2015 at 2:37 PM, Carin Meier <[email protected]> >>> wrote: >>> >>>> Wow. I didn't realize there was HTM.Java. That makes my Clojure >>>> interop very happy. I will have to investigate and explore. >>>> >>>> Thanks! >>>> >>>> On Fri, Oct 16, 2015 at 3:33 PM, Matthew Taylor <[email protected]> >>>> wrote: >>>> >>>>> 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 >>>>> >>>>> >>>> >>> >>> >>> -- >>> *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 >>> >> >> >
