I'm not sure if it will work, but I would try this:

1. Create a model that takes two fields, one float field (let's call it
"number") and one category field (let's call it "label").

2. Make "label" the predictedField.

3. For the sequences (1,2,3 => 'seq1') and (4,5,8 => 'seq2'), feed them to
the model as follows:

number, label
------------------
1, seq1
2, seq1
3, seq1
RESET SEQUENCE
4, seq2
5, seq2
8, seq2

4. Disable learning.

5. Feed the test sequence, and get predictions:

number, label
------------------
1.1,
1.9,
2.8,

6. Let us know if it worked!


On Thu, Apr 3, 2014 at 5:34 AM, Scheele, Manuel <
[email protected]> wrote:

> Hi Matt,
>
> Yes, that is what I want to do. I want to show the CLA a sequences with
> labels (call sequence 1,2,3 'seq1' and sequence 4,5,8 'seq2') and then show
> noisy sequences and see if it can assign the correct label to it ( assign
> label 'seq1' to sequence 1.1,1.9,2.8). Can you help me to find the correct
> "inferenceType" and "inferenceArgs" for the search_def.json for swarming?
>
> Regards,
> Manuel
> ________________________________________
> From: nupic [[email protected]] on behalf of Matthew Taylor
> [[email protected]]
> Sent: 02 April 2014 18:21
> To: NuPIC general mailing list.
> Subject: Re: [nupic-discuss] NuPIC Recognition Tasks
>
> Can you elaborate a bit? You want to do sequence recognition but not
> prediction. Does that mean you want to recognize the current sequences the
> CLA recognizes are occurring at any point in the data?
>
> ---------
> Matt Taylor
> OS Community Flag-Bearer
> Numenta
>
>
> On Tue, Apr 1, 2014 at 11:34 PM, Scheele, Manuel <
> [email protected]<mailto:[email protected]>>
> wrote:
> Hi All,
>
> When I want to do sequence recognition (i.e. no prediction, just
> recognition of the input) what should I set the "inferenceType" and
> "inferenceArgs" to in the search_def.json for swarming?
>
> Thanks for the help,
> Manuel
> _______________________________________________
> nupic mailing list
> [email protected]<mailto:[email protected]>
> http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org
>
> _______________________________________________
> nupic mailing list
> [email protected]
> http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org
>
_______________________________________________
nupic mailing list
[email protected]
http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org

Reply via email to