Thank you so much! It is very helpful to me.
- Mino 2015-07-28 5:52 GMT+09:00 Matthew Taylor <[email protected]>: > Hi Mino, > > > Please watch this video, I think it will be very informative for your > question: > > https://www.youtube.com/watch?v=QZBtaP_gcn0 > > I ask Subutai a question specifically related to your question at 55m > 32s : https://youtu.be/QZBtaP_gcn0?t=55m32s > --------- > Matt Taylor > OS Community Flag-Bearer > Numenta > > > On Sun, Jul 26, 2015 at 11:51 AM, Min-Oh Heo <[email protected]> wrote: > > Hello, > > > > > > > > I’m a newbie, here. This is my first posting. > > > > > > > > > > > > I’m interested in geospatial tracking demo. > > > > > > > > In some articles in the mail archives, the application seems to be one of > > very good examples of NuPIC to show its advantages. > > > > > > > > I’d like to test the applicability for ‘anomaly detection task’ and > > ‘next-step position prediction task’. > > > > > > > > But, I realized the next-step position prediction function is not > > implemented in the current demo program. > > > > > > > > > > So, I tried implementing geospatial predictor with the one hot-gym > > prediction code. > > > > > > On the way to do, I realized that the geospatial coordinate encoders in > > model configuration seems impossible to decode due to randomness and hash > > function. > > > > > > Could you please let me know what kind of approaches can be applied to > > implement the next-step position prediction with NuPIC.geospatial? > > > > > > > > > > > > Any suggestion would be very helpful. > > > > > > > > Thank you in advance! > > > > > > > > - Mino > >
