Agreed: https://github.com/numenta/nupic/issues/1461

It might also be cool to have a 3D Coordinate Encoder that does the
same thing for X,Y,Z coordinates (where Z is optional).
---------
Matt Taylor
OS Community Flag-Bearer
Numenta


On Fri, Oct 24, 2014 at 11:44 AM, Chetan Surpur <[email protected]> wrote:
> If people are finding that altitude is a parameter they care about, I
> suggest modifying the GeospatialCoordinateEncoder to also take in altitude.
> Then, for whoever wants a 2D geospatial encoder, they can just use a fixed
> altitude. I feel this is the cleanest approach.
>
>
>
> On Fri, Oct 24, 2014 at 2:02 AM, Fergal Byrne <[email protected]>
> wrote:
>>
>> Hi Cameron,
>>
>> Matt is correct: Chetan's coordinate encoders are the way to go. I'd
>> follow his method and extend the CoordinateEncoder to three dimensions -
>> perhaps calling it the GPSEncoder - with speed used to form a radius in 3D
>> space.
>>
>> Regards
>>
>> Fergal Byrne
>>
>> On Fri, Oct 24, 2014 at 7:02 AM, Matthew Taylor <[email protected]> wrote:
>>>
>>> Sounds awesome! Let's talk about data. Do you know what kind of data
>>> you'll be getting from these planes and ships? How many individual tracks
>>> are you talking about?
>>>
>>> We have two types of encoders currently for spatial data. The
>>> GeospatialCoordinateEncoder [1] is fed lat/lon coords and velocity, but it
>>> doesn't deal with altitude. It is a subclass of the CoordinateEncoder [2],
>>> which can be fed any array of coordinates and a radius. For more details on
>>> these encoders, see Chetan's video [3].
>>>
>>> If altitude data is important, I suggest you try using the
>>> CoordinateEncoder, and manually calculate a radius for the encoder using
>>> velocity.
>>>
>>> [1]
>>> https://github.com/numenta/nupic/blob/master/nupic/encoders/geospatial_coordinate.py
>>> [2]
>>> https://github.com/numenta/nupic/blob/master/nupic/encoders/coordinate.py
>>> [3] http://www.youtube.com/watch?v=KxxHo-FtKRo
>>>
>>> ---------
>>> Matt Taylor
>>> OS Community Flag-Bearer
>>> Numenta
>>>
>>> On Thu, Oct 23, 2014 at 6:14 PM, Cameron Hunt <[email protected]>
>>> wrote:
>>>>
>>>> I want to share a new project we're undertaking: integrating NuPIC into
>>>> our current environment where we do analysis on plane and ship traffic from
>>>> ADS-B and AIS messages. Integrating NuPIC is part of an overall rework of
>>>> our environment to shift to using Hadoop as our data persistence and
>>>> processing environment, and we're adding in Cesium/Geomesa for our
>>>> geospatial UI/UX, and Lab41's Dendrite/Titan distro for our graph data
>>>> UI/UX.
>>>>
>>>> Our goal would be to add anomaly scores generated by NuPIC into our user
>>>> review process. This means that for us it is as important to understand how
>>>> humans work with identified anomalies and feed their analysis back into the
>>>> algorithm.
>>>>
>>>> C4OE is a new non-profit that was created - in part - to better support
>>>> development of Open Source analytical software. So everything we develop 
>>>> (or
>>>> that we fund others to develop for us) will be released under an 
>>>> appropriate
>>>> Open Source license.
>>>>
>>>> --
>>>> Cameron Hunt
>>>> Director, Center For Open Exploration
>>>> Cell: 843.654.4708
>>>
>>>
>>
>>
>>
>> --
>>
>> Fergal Byrne, Brenter IT
>>
>> http://inbits.com - Better Living through Thoughtful Technology
>> http://ie.linkedin.com/in/fergbyrne/ - https://github.com/fergalbyrne
>>
>> Founder of Clortex: HTM in Clojure -
>> https://github.com/nupic-community/clortex
>>
>> Author, Real Machine Intelligence with Clortex and NuPIC
>> Read for free or buy the book at https://leanpub.com/realsmartmachines
>>
>> Speaking on Clortex and HTM/CLA at euroClojure Krakow, June 2014:
>> http://euroclojure.com/2014/
>> and at LambdaJam Chicago, July 2014: http://www.lambdajam.com
>>
>> e:[email protected] t:+353 83 4214179
>> Join the quest for Machine Intelligence at http://numenta.org
>> Formerly of Adnet [email protected] http://www.adnet.ie
>
>

Reply via email to