public void Test_Train_Correct()
{
var inputArray = new double[, , ]
{
{
{ 1, 3, 5, 8, 7, 4, 2, 8, 6 },
{ 4, 2, 3, 5, 1, 0, 1, 2, 7 },
{ 8, 3, 1, 2, 3, 6, 5, 8, 5 },
{ 7, 2, 4, 2, 9, 3, 6, 4, 6 },
{ 9, 8, 0, 8, 8, 3, 7, 1, 0 },
{ 9, 4, 1, 2, 2, 4, 7, 4, 3 },
{ 0, 5, 9, 1, 6, 4, 5, 4, 3 },
{ 3, 0, 5, 8, 6, 5, 7, 0, 1 },
{ 1, 4, 2, 3, 6, 1, 5, 4, 3 }
},
{
{ 4, 3, 5, 0, 1, 2, 4, 8, 9 },
{ 4, 8, 0, 1, 3, 4, 5, 6, 7 },
{ 4, 4, 2, 3, 6, 5, 4, 2, 1 },
{ 0, 7, 8, 6, 4, 4, 1, 0, 0 },
{ 1, 0, 1, 2, 4, 5, 6, 2, 3 },
{ 1, 2, 3, 7, 8, 8, 7, 4, 6 },
{ 0, 3, 3, 2, 1, 7, 2, 3, 0 },
{ 1, 8, 4, 4, 1, 2, 0, 0, 5 },
{ 7, 9, 9, 9, 1, 3, 1, 4, 6 }
},
{
{ 7, 1, 4, 3, 2, 1, 0, 4, 6 },
{ 5, 3, 1, 9, 7, 5, 3, 1, 1 },
{ 4, 2, 0, 8, 6, 4, 2, 0, 0 },
{ 7, 8, 7, 4, 3, 8, 0, 1, 4 },
{ 5, 2, 1, 3, 6, 5, 4, 7, 5 },
{ 1, 6, 2, 4, 2, 3, 7, 6, 1 },
{ 0, 3, 7, 1, 3, 0, 3, 1, 8 },
{ 6, 4, 8, 8, 9, 0, 1, 2, 1 },
{ 5, 6, 7, 8, 9, 0, 2, 4, 7 }
}
};
var filter1Array = new double[, , ]
{
{
{ 1, 4, 5 },
{ 0, 3, 4 },
{ 6, 8, 7 }
},
{
{ 4, 1, 0 },
{ 6, 3, 1 },
{ 5, 4, 2 }
},
{
{ 2, 2, 5 },
{ 3, 7, 4 },
{ 4, 1, 0 }
}
};
var filter2Array = new double[, , ]
{
{
{ 4, 7, 1 },
{ 3, 0, 2 },
{ 4, 2, 1 }
},
{
{ 3, 2, 2 },
{ 8, 3, 4 },
{ 7, 7, 1 }
},
{
{ 2, 3, 0 },
{ 4, 0, 7 },
{ 1, 5, 4 }
}
};
var expectedArray = new double[, , ]
{
{
{ 336, 320, 339, 402 },
{ 360, 402, 448, 266 },
{ 272, 377, 444, 353 },
{ 408, 486, 299, 253 }
},
{
{ 287, 312, 334, 336 },
{ 330, 335, 379, 317 },
{ 272, 300, 424, 352 },
{ 410, 481, 250, 248 }
}
};
var errorArray = new double[, , ]
{
{
{ 1, 2, 3, 4 },
{ 5, 6, 7, 8 },
{ 1, 2, 3, 4 },
{ 5, 6, 7, 8 }
},
{
{ 5, 4, 3, 2 },
{ 1, 2, 3, 4 },
{ 6, 5, 4, 3 },
{ 4, 3, 2, 1 }
}
};
var input = Converters.Convert(inputArray);
var filter1 = Converters.Convert(filter1Array);
var filter2 = Converters.Convert(filter2Array);
var expected = Converters.Convert(expectedArray);
var error = Converters.Convert(errorArray);
var layer = new ConvolutionalLayer
{
Stride = 2,
ActivationFunction = new ActivationFunctionMoq()
};
layer.Neurals.Add(new Filter {
Weights = filter1
});
layer.Neurals.Add(new Filter {
Weights = filter2
});
var newError = layer.Train(error, input, expected);
}