Accord in C++/cli

[sp]

Hi all,

I'm trying to use Accord.Net in c++/cli.

Did anyone try it ?

I have translated the simple example depicted in the getting started page into c++/cli code. (see the original program written in c# here: http://code.google.com/p/accord/wiki/GettingStarted)

It all seems okay, however I'm getting some annoying error message which saying: cannot convert parameter 2 from 'cli::array<Type,dimension> ^' to 'cli::array<Type,dimension> ^'

please see also the attached screen-shots.

I just don't get it since it's the same parameter...

Any idea ?

thanks in advanced!
[Please Excuse My English.:)]

[César]

Hi there,

 

I suspect you are getting this error because the SequentialMinimalOptimization constructor expects an int[] array as the second parameter. Please try changing the type of the _outputs array from cli::array<double,1> to cli::array<int,1>.

 

Please let me know if it works!

 

Best regards,

Cesar

[sp]

Cesar, thanks of your answer.

Unfortunately this didn't solve the problem, please see the attached:

(BTW, the input and not the output is the second parameter)

[César]

Hi,

 

Here is a working example. The problem was that the second parameter was expected to be a jagged array; completely overlooked that. Hope it helps this time.

 

Best regards,

Cesar

 

 

// cpp_test.cpp : main project file.

#include "stdafx.h"

using namespace cli;
using namespace System;
using namespace Accord::MachineLearning::VectorMachines;
using namespace Accord::MachineLearning::VectorMachines::Learning;
using namespace Accord::Math;
using namespace Accord::Statistics::Kernels;

int main(array<System::String ^> ^args)
{
    KernelSupportVectorMachine^ ksvm = gcnew KernelSupportVectorMachine(gcnew Gaussian(), 2);

    // Example XOR problem
    array<array<double>^>^ inputs =
    {
        gcnew array<double> { 0, 0 }, // 0 xor 0: 1 (label +1)
        gcnew array<double> { 0, 1 }, // 0 xor 1: 0 (label -1)
        gcnew array<double> { 1, 0 }, // 1 xor 0: 0 (label -1)
        gcnew array<double> { 1, 1 }  // 1 xor 1: 1 (label +1)
    };

    // Dichotomy SVM outputs should be given as [-1;+1]
    array<int>^ outputs =
    {
       1, -1, -1, 1
    };

    SequentialMinimalOptimization^ smo = gcnew SequentialMinimalOptimization(ksvm, inputs, outputs);

    double error = smo->Run();

    return 0;
}

 

[sp]

Working indeed!

thank you very much!