You should have a look at MultiEncoder.
https://github.com/numenta/nupic/blob/master/nupic/encoders/multi.py
It does basically what Chandan describes :)
On Jun 23, 2015 8:13 PM, "Chandan Maruthi" <[email protected]>
wrote:

> I havent played with hierarchy yet , But I assume something like this must
> be possible with hierarchy too.
>
> On Tue, Jun 23, 2015 at 10:59 AM, Chandan Maruthi <
> [email protected]> wrote:
>
>> Jose,
>> Theoretically this is possible. When a prediction is made, you are
>> basically referring to a state  that the CLA has seen before in the given
>> context occurring after the current step. If this state is a combination
>> off multiple inputs then the state can be resolved into multiple inputs
>> that  represent the state.
>>
>> To achieve this out of the box. May be you can write your own encoder
>> that creates an output where every x bits represent a given input.  As you
>> have mentioned in your comment .
>>
>> On Tue, Jun 23, 2015 at 10:19 AM, Jose Luis Contreras Santos <
>> [email protected]> wrote:
>>
>>> Hello all!
>>>
>>> I would like to ask you a question: do you know if it is possible to
>>> make predictions about more than one field at the same time? If not, I
>>> would assume that is due to implementation, as theoretically it should be
>>> possible, am I right?
>>>
>>> Thanks for your help!
>>>
>>> Jose Luis
>>>
>>> PS: However, I imagine that it could be possible to group variables into
>>> states and predict states instead of individual values for variables - as
>>> long as they have a limited number of possible values
>>>
>>
>>
>>
>> --
>> Regards
>> Chandan Maruthi
>>
>>
>
>
> --
> Regards
> Chandan Maruthi
>
>

Reply via email to