mic::neural_nets::unit_tests Namespace Reference

Classes
class	Conv2x2x2Filter2x1x1s1Double
	Test Fixture - layer of input size 2x2x2 and with filter bank of 2 filters of size 1x1 with stride 1, double. Math example taken from my own calculations;) More...
class	Conv3x3x2Filter3x2x2s1Float
	Test Fixture - layer of input size 3x3x2 and with filter bank of 3 filters of size 2x2 with stride 1, floats. Math example taken from my whiteboard;) More...
class	Conv4x4x1Filter1x2x2s2Float
	Test Fixture - layer of input size 4x4x1 and with filter bank of 1 filters of size 2x2 with stride 2, floats. Math example taken from my own YET ANOTHER calculations! ech! More...
class	Conv4x4x1Filter3x1x1s3Double
	Test Fixture - layer of input size 4x4x1 and with filter bank of 3 filters of size 1x1 with stride 3, double. Math example taken from my own calculations;) More...
class	Conv5x5x1Filter1x3x3s1Float
	Test Fixture - layer of input size 5x5x1 and with filter bank of 1 filter of size 3x3 with stride 1 (floats). Math example taken from: https://ujjwalkarn.me/2016/08/11/intuitive-explanation-convnets/. More...
class	Conv5x5x1Filter1x2x2s3Float
	Test Fixture - layer of input size 5x5x1 and with filter bank of 1 filter of size 2x2 with stride 3 (float). Math example taken from: https://ujjwalkarn.me/2016/08/11/intuitive-explanation-convnets/. More...
class	Conv7x7x3Filter3x3x3s2Float
	Test Fixture - layer of input size 7x7x3 and with filter bank of 2 filters of 3x3 with stride 2 (floats). Math example taken from: http://cs231n.github.io/convolutional-networks/. More...
class	Conv5x6x1Filter1x4x4s1Float
	Test Fixture - layer of input size 5x6x1 and with filter bank of 1 filter of size 4x4 with stride 1, floats. Math example taken from: http://soumith.ch/ex/pages/2014/08/07/why-rotate-weights-convolution-gradient/. More...
class	Conv28x28x1Filter2x28x28s1Double
	Test Fixture - layer of input size 28x28x1 and with filter bank of 2 filters of size 28x28 with stride 1, double. More...
class	Conv8x8x1Filter2x4x4s4Double
	Test Fixture - layer of input size 8x8x1 and with filter bank of 2 filters of size 4x4 with stride 4, double. More...
class	Linear1x1Float
	Test Fixture - layer of size 1x1, floats, sets W[0] = 1.0 and b[0] = 1.0. More...
class	Linear5x2Float
	Test Fixture - layer of size 5x2, floats. More...
class	Linear2x3Float
	Test Fixture - layer of size 2x3, floats, sets all internal and external values. More...
class	Linear2x3Double
	Test Fixture - layer of size 2x3, doubles, sets all internal and external values. More...
class	Linear50x100Double
	Test Fixture - layer of size 50x100, doubles, randomly sets all internal and external values required for numerical gradient verification. More...
class	Simple2LayerRegressionNN
	Test Fixture - simple ff net with 2 layers. More...
class	Tutorial2LayerNN
	Test Fixture - feed-forward net with 2 layers. A "formalized" example from a step-by-step tutorial: https://mattmazur.com/2015/03/17/a-step-by-step-backpropagation-example/. More...

Functions
	TEST_F (Conv2x2x2Filter2x1x1s1Double, NumericalGradientCheck)
	Numerical gradient test of all parameters for layer of input size 2x2x2 and with filter bank of 2 filters of size 1x1 with stride 1. More...
	TEST_F (Conv28x28x1Filter2x28x28s1Double, NumericalGradientCheck)
	Numerical gradient test of all parameters for layer of input size 28x28x1 and with filter bank of 2 filters of size 28x28 with stride 1, double. More...
	TEST_F (Conv28x28x1Filter2x28x28s1Double, Convergence)
	TEST (Convolutions, LayerDimensions)
	TEST_F (Conv2x2x2Filter2x1x1s1Double, Forward)
	TEST_F (Conv2x2x2Filter2x1x1s1Double, Backward)
	TEST_F (Conv3x3x2Filter3x2x2s1Float, Forward)
	TEST_F (Conv4x4x1Filter1x2x2s2Float, Forward)
	TEST_F (Conv4x4x1Filter1x2x2s2Float, Backward)
	TEST_F (Conv4x4x1Filter3x1x1s3Double, Forward)
	TEST_F (Conv4x4x1Filter3x1x1s3Double, Backward)
	TEST_F (Conv5x5x1Filter1x3x3s1Float, Dimensions)
	TEST_F (Conv5x5x1Filter1x3x3s1Float, Forward)
	TEST_F (Conv5x5x1Filter1x2x2s3Float, Forward)
	TEST_F (Conv5x5x1Filter1x2x2s3Float, Backward)
	TEST_F (Conv5x6x1Filter1x4x4s1Float, Forward)
	TEST_F (Conv7x7x3Filter3x3x3s2Float, Forward)
	TEST_F (Linear5x2Float, WbInitialization)
	Makes sure that the layer is properly initialized - initial W weights must be non zero and b must be zero. More...
	TEST_F (Linear5x2Float, WIsNaN)
	Makes sure that the layer is properly initialized - all W are numbers! More...
	TEST_F (Linear5x2Float, WIsNotInf)
	Makes sure that the layer is properly initialized - all W are finite. More...
	TEST_F (Linear5x2Float, WAreDifferent)
	Makes sure that the layer is properly initialized and all W are different. More...
	TEST_F (Linear1x1Float, Forward_y)
	Makes sure that the layer calculates y = w*x + b, size of layer: is 1x1. More...
	TEST_F (Linear2x3Float, Forward_y)
	Makes sure that the layer calculates y = w*x + b, size of layer: is 2x3. More...
	TEST (LinearStacked1x2x3Float, Forward_y)
	Makes sure that the two stacked layers will return right result. More...
	TEST (Linear2x1Float, Backward_dx)
	Tests backward pass in the y = w*x + b, size of layer: is 2x1. More...
	TEST_F (Linear2x3Float, Backward_dx)
	Tests backward pass in the y = w*x + b (dx gradient), size of layer: is 2x3. More...
	TEST_F (Linear2x3Float, Backward_dWdb)
	Tests gradients dW and db, size of layer is 2x3. More...
	TEST_F (Linear2x3Double, NumericalGradientCheck_dW)
	Numerical gradient test dW, size of layer is 2x3. More...
	TEST_F (Linear2x3Double, NumericalGradientCheck_db)
	Numerical gradient test db, size of layer is 2x3. More...
	TEST_F (Linear2x3Double, NumericalGradientCheck_dx)
	Numerical gradient test dx, size of layer is 2x3. More...
	TEST_F (Linear50x100Double, NumericalGradientCheck_dW)
	Numerical gradient test dW, size of layer is 50x100. More...
	TEST_F (Linear50x100Double, NumericalGradientCheck_db)
	Numerical gradient test db, size of layer is 50x100. More...
	TEST_F (Linear50x100Double, NumericalGradientCheck_dx)
	Numerical gradient test dx, size of layer is 50x100. More...
	TEST_F (Simple2LayerRegressionNN, Dimensions)
	TEST_F (Simple2LayerRegressionNN, BatchResize)
	TEST_F (Simple2LayerRegressionNN, Serialization)
	TEST_F (Tutorial2LayerNN, BackpropagationSingleStep)
	TEST_F (Tutorial2LayerNN, TrainSingleStep)

Function Documentation

mic::neural_nets::unit_tests::TEST	(	Convolutions	,
		LayerDimensions
	)

Checks numbers of receptive fields for different strides.

Author

tkornuta

Definition at line 33 of file Convolution_tests.cpp.

References mic::mlnn::Layer< eT >::output_height, mic::mlnn::Layer< eT >::output_width, and mic::mlnn::Layer< eT >::s.

mic::neural_nets::unit_tests::TEST	(	LinearStacked1x2x3Float	,
		Forward_y
	)

Makes sure that the two stacked layers will return right result.

Author

tkornuta

Definition at line 101 of file Linear_tests.cpp.

References mic::mlnn::fully_connected::Linear< eT >::forward(), and mic::mlnn::Layer< eT >::p.

Here is the call graph for this function:

mic::neural_nets::unit_tests::TEST	(	Linear2x1Float	,
		Backward_dx
	)

Tests backward pass in the y = w*x + b, size of layer: is 2x1.

Author

tkornuta

Definition at line 127 of file Linear_tests.cpp.

References mic::mlnn::fully_connected::Linear< eT >::backward(), and mic::mlnn::Layer< eT >::p.

Here is the call graph for this function:

mic::neural_nets::unit_tests::TEST_F	(	Simple2LayerRegressionNN	,
		Dimensions
	)

Tests the dimensionality of nn.

Definition at line 32 of file MultiLayerNeuralNetworkTests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear5x2Float	,
		WbInitialization
	)

Makes sure that the layer is properly initialized - initial W weights must be non zero and b must be zero.

Author

tkornuta

Definition at line 33 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv2x2x2Filter2x1x1s1Double	,
		NumericalGradientCheck
	)

Numerical gradient test of all parameters for layer of input size 2x2x2 and with filter bank of 2 filters of size 1x1 with stride 1.

Author

tkornuta

Definition at line 34 of file Convolution_convergence_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear5x2Float	,
		WIsNaN
	)

Makes sure that the layer is properly initialized - all W are numbers!

Author

tkornuta

Definition at line 45 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Simple2LayerRegressionNN	,
		BatchResize
	)

Tests the batch resizing.

Definition at line 51 of file MultiLayerNeuralNetworkTests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear5x2Float	,
		WIsNotInf
	)

Makes sure that the layer is properly initialized - all W are finite.

Author

tkornuta

Definition at line 54 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear5x2Float	,
		WAreDifferent
	)

Makes sure that the layer is properly initialized and all W are different.

Author

tkornuta

Definition at line 64 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Simple2LayerRegressionNN	,
		Serialization
	)

Tests squared error loss function on vectors with four floats.

Definition at line 68 of file MultiLayerNeuralNetworkTests.cpp.

References fileName, and mic::mlnn::MultiLayerNeuralNetwork< eT >::load().

Here is the call graph for this function:

mic::neural_nets::unit_tests::TEST_F	(	Linear1x1Float	,
		Forward_y
	)

Makes sure that the layer calculates y = w*x + b, size of layer: is 1x1.

Author

tkornuta

Definition at line 76 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv2x2x2Filter2x1x1s1Double	,
		Forward
	)

Checks whether the forward is working for layer of input size 2x2x2 and with filter bank of 2 filters of size 1x1 with stride 1.

Author

tkornuta

Definition at line 78 of file Convolution_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv28x28x1Filter2x28x28s1Double	,
		NumericalGradientCheck
	)

Numerical gradient test of all parameters for layer of input size 28x28x1 and with filter bank of 2 filters of size 28x28 with stride 1, double.

Author

tkornuta

Definition at line 84 of file Convolution_convergence_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear2x3Float	,
		Forward_y
	)

Makes sure that the layer calculates y = w*x + b, size of layer: is 2x3.

Author

tkornuta

Definition at line 90 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Tutorial2LayerNN	,
		BackpropagationSingleStep
	)

Tests a single iteration of a backpropagation algorithm.

Definition at line 100 of file MultiLayerNeuralNetworkTests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv2x2x2Filter2x1x1s1Double	,
		Backward
	)

Checks whether the backward pass is working for layer of input size 2x2x2 and with filter bank of 2 filters of size 1x1 with stride 1.

Author

tkornuta

Definition at line 102 of file Convolution_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv28x28x1Filter2x28x28s1Double	,
		Convergence
	)

Checks convergence of layer of input size 28x28x1 and with filter bank of 2 filters of size 28x28 with stride 1, double.

Author

tkornuta

Definition at line 134 of file Convolution_convergence_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv3x3x2Filter3x2x2s1Float	,
		Forward
	)

Checks whether the forward is working for layer of input size 3x3x2 and with filter bank of 3 filters of size 2x2 with stride 1.

Author

tkornuta

Definition at line 139 of file Convolution_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear2x3Float	,
		Backward_dx
	)

Tests backward pass in the y = w*x + b (dx gradient), size of layer: is 2x3.

Author

tkornuta

Definition at line 146 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear2x3Float	,
		Backward_dWdb
	)

Tests gradients dW and db, size of layer is 2x3.

Author

tkornuta

Definition at line 158 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv4x4x1Filter1x2x2s2Float	,
		Forward
	)

Checks whether the forward is working for layer of input size 4x4x1 and with filter bank of 1 filters of size 2x2 with stride 2.

Author

tkornuta

Definition at line 158 of file Convolution_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Tutorial2LayerNN	,
		TrainSingleStep
	)

Tests a single iteration of a backpropagation algorithm.

Definition at line 165 of file MultiLayerNeuralNetworkTests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv4x4x1Filter1x2x2s2Float	,
		Backward
	)

Checks whether the backward gradient pass is working for layer of input size 4x4x1 and with filter bank of 1 filters of size 2x2 with stride 2.

Author

tkornuta

Definition at line 174 of file Convolution_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear2x3Double	,
		NumericalGradientCheck_dW
	)

Numerical gradient test dW, size of layer is 2x3.

Author

tkornuta

Definition at line 182 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv4x4x1Filter3x1x1s3Double	,
		Forward
	)

Checks whether the forward is working for layer of input size 4x4x1 and with filter bank of 3 filters of size 1x1 with stride 3, double.

Author

tkornuta

Definition at line 200 of file Convolution_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear2x3Double	,
		NumericalGradientCheck_db
	)

Numerical gradient test db, size of layer is 2x3.

Author

tkornuta

Definition at line 208 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv4x4x1Filter3x1x1s3Double	,
		Backward
	)

Checks whether the backward gradient pass is working for layer of input size 4x4x1 and with filter bank of 3 filters of size 1x1 with stride 3, double.

Author

tkornuta

Definition at line 216 of file Convolution_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear2x3Double	,
		NumericalGradientCheck_dx
	)

Numerical gradient test dx, size of layer is 2x3.

Author

tkornuta

Definition at line 232 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv5x5x1Filter1x3x3s1Float	,
		Dimensions
	)

Checks whether dimensions of xs, outputs and filters are ok. Convolutional dimensions are nicely explained in this lecture: http://cs231n.github.io/convolutional-networks/

Author

tkornuta

Definition at line 243 of file Convolution_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear50x100Double	,
		NumericalGradientCheck_dW
	)

Numerical gradient test dW, size of layer is 50x100.

Author

tkornuta

Definition at line 256 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv5x5x1Filter1x3x3s1Float	,
		Forward
	)

Checks whether the forward is working for layer of input size 5x5x1 and with filter bank of 1 filter of size 3x3 with stride 1.

Author

tkornuta

Definition at line 268 of file Convolution_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv5x5x1Filter1x2x2s3Float	,
		Forward
	)

Checks whether the forward is working for layer of input size 5x5x1 and with filter bank of 1 filter of size 2x2 with stride 3 (float).

Author

tkornuta

Definition at line 284 of file Convolution_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear50x100Double	,
		NumericalGradientCheck_db
	)

Numerical gradient test db, size of layer is 50x100.

Author

tkornuta

Definition at line 285 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv5x5x1Filter1x2x2s3Float	,
		Backward
	)

Checks whether the backward gradient pass is working for layer of input size 4x4x1 and with filter bank of 3 filters of size 1x1 with stride 3, double.

Author

tkornuta

Definition at line 300 of file Convolution_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Linear50x100Double	,
		NumericalGradientCheck_dx
	)

Numerical gradient test dx, size of layer is 50x100.

Author

tkornuta

Definition at line 313 of file Linear_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv5x6x1Filter1x4x4s1Float	,
		Forward
	)

Checks whether the forward is working for layer of input size 5x6x1 and with filter bank of 1 filter of size 4x4 with stride 1.

Author

tkornuta

Definition at line 324 of file Convolution_tests.cpp.

mic::neural_nets::unit_tests::TEST_F	(	Conv7x7x3Filter3x3x3s2Float	,
		Forward
	)

Checks whether the forward is working for layer of input size 7x7x3 and with filter bank of 3 filters of 3x3 size with stride 2.

Author

tkornuta

Definition at line 342 of file Convolution_tests.cpp.