I think i'm not explaining myself correctly. This could be due to the
fact that I'm new to the vocabulary of machine learning.
Let's take a binary regression. It take X as an input and gives 0 or
1. If I want to train such regression I need a matrix Xdata which
could represents 'm' pictures of cats and Ydata which is a vector of
length 'm' constituted of 0 and 1 depending if the associated image is
a cat or not. To train such a network we need both Xdata and Ydata. We
then use gradient descent to find the weights to get the best
approximate of f(X)->Y
In the case of the HPX smart executors, the regression takes as an
input static and dynamic features and gives as an output the optimal
policy(sequential or parallel). In that case, Xdata would represents
different static and dynamic features for 'm' tests. But to train the
network, you also need Ydata which,in that case, is the optimal policy
for each of the 'm' training examples.
From what you tell me it seems that we don't have Ydata so we don't
know the optimal policies for our training example. If it's the case,
I simply don't understand how you can apply a learning algorithm. My
analogy would be that you can't train a regression to recognize cat
pictures if you don't tell it which ones of the pictures you it train
with are cats and which aren't.
Thank you very much.
Patrick Diehl <patrickdie...@gmail.com> a écrit :
> Hi Gabriel,
>> I wanted to know how you are able to find
>>> the Y vector which is the expected output of the regression for the
>>> input data? Am I understanding correctly?
> You have to run the code with x_1,x_2,...,x_n and measure the y values
> for those input values.
> On 22/02/18 05:41 PM, Gabriel Laberge wrote:
>> I understand but what I meant is that to train a logistic regression
>> you need data that is made of a X matrix which represents the input
>> data. Lets say we have 'm' examples. We also need a Y vector which
>> represent the expected outputs for each of the 'm' training example.
>> The training process is simply fitting the weights of the regression
>> so the output of the regression is almost always outputs the right Y
>> when you give an input X. I wanted to know how you are able to find
>> the Y vector which is the expected output of the regression for the
>> input data? Am I understanding correctly?
>> Zahra Khatami <z.khatam...@gmail.com> a écrit :
>>> I am not sure if I understand your concern correctly. The optimal
>>> parameters ( chunk size, preferching distance or policies) shouldn’t be
>>> found before training data. They are found for each of the hpx loops at
>>> runtime based on the loop static and dynamic parameters. That’s a main goal
>>> of this research. The candidates of these optimal parameters are chosen
>>> when training model. Then the optimal one will be selected between them at
>>> runtime, which may be different for each loop with different parameters.
>>> On Tue, Feb 20, 2018 at 7:51 AM Gabriel Laberge
>>>> I had a questions on the way data was generated in order to train the
>>>> logistics regressions models talked about in 
>>>> For each of the training examples, the optimal execution
>>>> policies,chunk sizes and prefetching distance had to be found before
>>>> the training process in order to have good data. I wonder if the
>>>> optimal parameters for the training examples were found by trial and
>>>> error or if there is another technique.
>>>> Thank you..
>>>> hpx-users mailing list
>>> Best Regards, Zahra Khatami | PhD Student Center for Computation &
>>> Technology (CCT) School of Electrical Engineering & Computer Science
>>> Louisiana State University 2027 Digital Media Center (DMC) Baton Rouge, LA
>> hpx-users mailing list
> hpx-users mailing list
hpx-users mailing list