msecnn_raulkviana.msecnn.LossFunctionMSE Class Reference

Inheritance diagram for msecnn_raulkviana.msecnn.LossFunctionMSE:

Public Member Functions
	__init__ (self, use_mod_cross_entropy=True, beta=1)
	get_proportion_CUs (self, labels)
	This function returns the proportion of CU's for all the types of split mode.
	get_min_RDs (self, RDs)
	Obtain the lowest value that isnt zero from RDs tensor.
	remove_values_lower (self, tensor, max_val, subst_val)
	Remove values from tensor that are lower than a given value.
	remove_inf_values (self, tensor)
	Remove values from tensor that are inf and make them zeros.
	remove_zero (self, RDs)
	Substitutes the zeros values for big RD values.
	remove_values_above (self, RDs, max_val)
	Substitutes values above MAX_RD for the MAX_RD.
	forward (self, pred, labels, RD)
	This function implements the loss function.

Public Attributes
	beta
	use_mod_cross_entropy
	MAX_RD
	MAX_LOSS
	last_loss
	last_pred
	last_RD

Constructor & Destructor Documentation

__init__()

msecnn_raulkviana.msecnn.LossFunctionMSE.__init__	(		self,
			use_mod_cross_entropy = True,
			beta = 1
	)

Member Function Documentation

forward()

msecnn_raulkviana.msecnn.LossFunctionMSE.forward	(		self,
			pred,
			labels,
			RD
	)

This function implements the loss function.

Parameters

[in]	pred	Predictions made by the model
[in]	labels	Ground-truth tensor
[in]	RD	Rate distortion tensor
[out]	loss	Vector of raw predictions that a classification model generates

get_min_RDs()

msecnn_raulkviana.msecnn.LossFunctionMSE.get_min_RDs	(		self,
			RDs
	)

Obtain the lowest value that isnt zero from RDs tensor.

Parameters

[in]	RDs	Tensor with RDs
[out]	min_RD	Lowest value of RDs that isnt zero

get_proportion_CUs()

msecnn_raulkviana.msecnn.LossFunctionMSE.get_proportion_CUs	(		self,
			labels
	)

This function returns the proportion of CU's for all the types of split mode.

Parameters

[in]	labels	Ground truth tensor
[out]	p_m	Tensor with the proportions

remove_inf_values()

msecnn_raulkviana.msecnn.LossFunctionMSE.remove_inf_values	(		self,
			tensor
	)

Remove values from tensor that are inf and make them zeros.

Parameters

[in]	tensor	Tensor with values
[out]	tensor	New tensor

remove_values_above()

msecnn_raulkviana.msecnn.LossFunctionMSE.remove_values_above	(		self,
			RDs,
			max_val
	)

Substitutes values above MAX_RD for the MAX_RD.

Parameters

[in]	RDs	Tensor with RDs
[in]	max_val	Max value to add
[out]	clone_RDs	Tensor With Max values added

remove_values_lower()

msecnn_raulkviana.msecnn.LossFunctionMSE.remove_values_lower	(		self,
			tensor,
			max_val,
			subst_val
	)

Remove values from tensor that are lower than a given value.

Parameters

[in]	tensor	Tensor with values
[in]	max_val	Threshold val
[in]	subst_val	Max value to replace the others
[out]	tensor	New tensor

remove_zero()

msecnn_raulkviana.msecnn.LossFunctionMSE.remove_zero	(		self,
			RDs
	)

Substitutes the zeros values for big RD values.

Parameters

[in]	RDs	Tensor with RDs
[out]	RDs	Tensor With Max values added

Member Data Documentation

beta

msecnn_raulkviana.msecnn.LossFunctionMSE.beta

last_loss

msecnn_raulkviana.msecnn.LossFunctionMSE.last_loss

last_pred

msecnn_raulkviana.msecnn.LossFunctionMSE.last_pred

last_RD

msecnn_raulkviana.msecnn.LossFunctionMSE.last_RD

MAX_LOSS

msecnn_raulkviana.msecnn.LossFunctionMSE.MAX_LOSS

MAX_RD

msecnn_raulkviana.msecnn.LossFunctionMSE.MAX_RD

use_mod_cross_entropy

msecnn_raulkviana.msecnn.LossFunctionMSE.use_mod_cross_entropy

---

The documentation for this class was generated from the following file:

• src/msecnn_raulkviana/msecnn.py