2013/9/23 Fred Baba :
> I have a python wrapper that allows me to export datasets to Python for
> analysis. But since my features are computed in C++, I'd like to compute
> predictions in C++ as well. Then I would simply set the coefficients at
> startup (v.first) and update the feature values cont
@Fred. It seems very interesting. Could you please provide more information
on how you achieved this?
On Mon, Sep 23, 2013 at 5:46 PM, Fred Mailhot wrote:
> FYI, I've used sklearn's LogisticRegression in an online/real-time text
> classification app without having to dig into the internals and g
2013/9/23 Fred Baba :
> Thanks, Olivier. The application involves continuous classification of
> real-time input features. So I only make on prediction at a time. My
> intention would be to observe the structure of the fitted model and then
> optimize the prediction function by hand in c++. For ins
Thanks, Olivier. The application involves continuous classification of
real-time input features. So I only make on prediction at a time. My
intention would be to observe the structure of the fitted model and then
optimize the prediction function by hand in c++. For instance, a linear
classifier wou
2013/9/23 abhishek :
> @Fred. It seems very interesting. Could you please provide more information
> on how you achieved this?
Using the default sklearn API (Pipeline of TfidfVectorizer +
LinearSVC) should work to provide ms-level prediction times. I think
there is nothing special to do in this c
I have a python wrapper that allows me to export datasets to Python for
analysis. But since my features are computed in C++, I'd like to compute
predictions in C++ as well. Then I would simply set the coefficients at
startup (v.first) and update the feature values continuously (v.second).
The entir
>2013/9/23 Fred Baba :
> System performance is currently on the order of ~1us, so Python overhead
> would be unacceptable. For SVM, I'll extract the support vectors and
> investigate using libSVM directly, as per federico vaggi's advice. +1 for
> PMML support at some point down the road. Thanks for
FYI, I've used sklearn's LogisticRegression in an online/real-time text
classification app without having to dig into the internals and gotten
~2.5ms response time (including vectorizing; vocab size ~200k).
On 23 September 2013 06:37, Peter Prettenhofer wrote:
> We don't have a PMML interface y
System performance is currently on the order of ~1us, so Python overhead
would be unacceptable. For SVM, I'll extract the support vectors and
investigate using libSVM directly, as per federico vaggi's advice. +1 for PMML
support at some point down the road. Thanks for the quick responses.
On Mon,
What are your requirements Fred ? in terms of maximum execution time ?
Predicting could be quite "fast" actually (think milliseconds) but depends
on your model and nber of features.
Timing a prediction code with timeit [1] should give you numbers for your
specific use case.
HTH,
[1] http://docs
As a generic answer, no. You'll have to write the equivalent of the
predict method yourself given the internal state (or, parameters) of the
trained classifier. For classifiers like SVM you can use libSVM directly
and avoid the Python wrapper.
On Mon, Sep 23, 2013 at 3:28 PM, Fred Baba wrote:
We don't have a PMML interface yet [1] - so you need to write custom code
to extract internal state each individual classifier.
What do you mean by performance critical (<1ms, <<1ms)? Do you make
predictions per sample or can you buffer samples and make predictions for
batches?
In general, what ki
I'd like to use classifiers trained via sklearn in a real-time application,
performance critical application. How do I access the internal
representation of trained classifiers?
For linear classifiers/regressions, I can simply store the coefficients and
generate the linear combination myself. For
13 matches
Mail list logo
