Sorry, I sent that incomplete (and this obviously remains untested):
class ToMultiInput(TransformerMixin, BaseEstimator):
def fit(self, shapes):
self.shapes = shapes
def transform(self, X):
shape_sizes = [np.prod(shape) for shape in self.shapes]
offsets = np.cumsum([0] + shape_sizes)
return [X[start:stop].reshape(shape)
for start, stop, shape
in zip(offsets, offsets[1:], self.shapes)]
tmi = ToMultiInput([single.shape for single in train_input])
train_input = np.hstack([single.reshape(single.shape[0], -1)
for single in train_input])
GridSearchCV(make_pipeline(tmi, my_predictor), ...)
On 1 May 2017 at 13:19, Joel Nothman <joel.noth...@gmail.com> wrote:
> Unless I'm mistaken about what we're looking at, you could use something
> like:
>
> class ToMultiInput(TransformerMixin, BaseEstimator):
> def fit(self, shapes):
> self.shapes = shapes
> def transform(self, X):
> return [X.]
>
> tmi = ToMultiInput([single.shape for single in train_input])
> # this assumes that train_input is a sequence of ndarrays with the same
> first dimension:
> train_input = np.hstack([single.reshape(single.shape[0], -1)
> for single in train_input])
>
> GridSearchCV(make_pipeline(tmi, my_predictor), ...)
>
>
> On 1 May 2017 at 11:45, Carlton Banks <nofl...@gmail.com> wrote:
>
>> How … batchsize could also be 1, I’ve just stored it like that.
>>
>> But how do reshape me data to be a matrix.. thats the big question.. is
>> possible?
>>
>> Den 1. maj 2017 kl. 02.21 skrev Joel Nothman <joel.noth...@gmail.com>:
>>
>> Do each of your 33 inputs have a batch of size 100? If you reshape your
>> data so that it all fits in one matrix, and then split it back out into its
>> 33 components as the first transformation in a Pipeline, there should be no
>> problem.
>>
>> On 1 May 2017 at 10:17, Joel Nothman <joel.noth...@gmail.com> wrote:
>>
>>> Sorry, I don't know enough about keras and its terminology.
>>>
>>> Scikit-learn usually limits itself to datasets where features and
>>> targets are a rectangular matrix.
>>>
>>> But grid search and other model selection tools should allow data of
>>> other shapes as long as they can be indexed on the first axis. You may be
>>> best off, however, getting support from the Keras folks.
>>>
>>> On 30 April 2017 at 23:23, Carlton Banks <nofl...@gmail.com> wrote:
>>>
>>>> It seems like scikit-learn is not able to handle network with multiple
>>>> inputs.
>>>> Keras documentation states:
>>>>
>>>> You can use Sequential Keras models (*single-input only*) as part of
>>>> your Scikit-Learn workflow via the wrappers found at
>>>> keras.wrappers.scikit_learn.py.
>>>> But besides what the wrapper can do.. can scikit-learn really not
>>>> handle multiple inputs?..
>>>>
>>>>
>>>> Den 30. apr. 2017 kl. 14.18 skrev Carlton Banks <nofl...@gmail.com>:
>>>>
>>>> The shapes are
>>>>
>>>> print len(train_input)print train_input[0].shapeprint train_output.shape
>>>> 33(100, 8, 45, 3)(100, 1, 145)
>>>>
>>>>
>>>> 100 is the batch-size..
>>>>
>>>> Den 30. apr. 2017 kl. 12.57 skrev Joel Nothman <joel.noth...@gmail.com
>>>> >:
>>>>
>>>> Scikit-learn should accept a list as X to grid search and index it just
>>>> fine. So I'm not sure that constraint applies to Grid Search
>>>>
>>>> On 30 April 2017 at 20:11, Julio Antonio Soto de Vicente <
>>>> ju...@esbet.es> wrote:
>>>>
>>>>> Tbh I've never tried, but I would say that te current sklearn API does
>>>>> not support multi-input data...
>>>>>
>>>>> El 30 abr 2017, a las 12:02, Joel Nothman <joel.noth...@gmail.com>
>>>>> escribió:
>>>>>
>>>>> What are the shapes of train_input and train_output?
>>>>>
>>>>> On 30 April 2017 at 12:59, Carlton Banks <nofl...@gmail.com> wrote:
>>>>>
>>>>>> I am currently trying to run some gridsearchCV on a keras model which
>>>>>> has multiple inputs.
>>>>>> The inputs is stored in a list in which each entry in the list is a
>>>>>> input for a specific channel.
>>>>>>
>>>>>>
>>>>>> Here is my model and how i use the gridsearch.
>>>>>>
>>>>>> https://pastebin.com/GMKH1L80
>>>>>>
>>>>>> The error i am getting is:
>>>>>>
>>>>>> https://pastebin.com/A3cB0rMv
>>>>>>
>>>>>> Any idea how i can resolve this?
>>>>>>
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> scikit-learn mailing list
>>>>>> scikit-learn@python.org
>>>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>>>>
>>>>>>
>>>>> _______________________________________________
>>>>> scikit-learn mailing list
>>>>> scikit-learn@python.org
>>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> scikit-learn mailing list
>>>>> scikit-learn@python.org
>>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>>>
>>>>>
>>>> _______________________________________________
>>>> scikit-learn mailing list
>>>> scikit-learn@python.org
>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> scikit-learn mailing list
>>>> scikit-learn@python.org
>>>> https://mail.python.org/mailman/listinfo/scikit-learn
>>>>
>>>>
>>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn@python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
>>
>>
>> _______________________________________________
>> scikit-learn mailing list
>> scikit-learn@python.org
>> https://mail.python.org/mailman/listinfo/scikit-learn
>>
>>
>
_______________________________________________
scikit-learn mailing list
scikit-learn@python.org
https://mail.python.org/mailman/listinfo/scikit-learn
