Ggtcf

How is it possible to have an ability score that is less than 3?

How much RAM could one put in a typical 80386 setup?

How long does it take to type this?

Why are only specific transaction types accepted into the mempool?

Copenhagen passport control - US citizen

Work Breakdown with Tikz

When blogging recipes, how can I support both readers who want the narrative/journey and ones who want the printer-friendly recipe?

Why is "Reports" in sentence down without "The"

How old can references or sources in a thesis be?

What is the offset in a seaplane's hull?

What do you call a Matrix-like slowdown and camera movement effect?

How do we improve the relationship with a client software team that performs poorly and is becoming less collaborative?

Shell script can be run only with sh command

I probably found a bug with the sudo apt install function

Are there any consumables that function as addictive (psychedelic) drugs?

How to report a triplet of septets in NMR tabulation?

Prevent a directory in /tmp from being deleted

What do you call something that goes against the spirit of the law, but is legal when interpreting the law to the letter?

Why are 150k or 200k jobs considered good when there are 300k+ births a month?

What makes Graph invariants so useful/important?

How do you conduct xenoanthropology after first contact?

What is the command to reset a PC without deleting any files

Is Social Media Science Fiction?

Circuitry of TV splitters

Gradient calculation within custom loss functions in Tensorflow/Keras

Make a custom loss function in kerasKeras/Tensorflow custom loss functionImplementing a custom loss function in KerasMake a custom loss function in kerasKeras custom loss function: Accessing current input patternWhy doesn't Keras need the gradient of a custom loss function?TensorFlow session inside Keras custom loss functionKeras: averaging different loss functionsKeras custom loss, having error in batch size in y_true and y_predUnderstanding Keras Loss Function Inputs in Reinforcement LearningKeras Custom Loss function to detect if an image is in a quadrant

.everyoneloves__top-leaderboard:empty,.everyoneloves__mid-leaderboard:empty,.everyoneloves__bot-mid-leaderboard:empty height:90px;width:728px;box-sizing:border-box;

0

I'm trying to build a custom loss function which utilizes the gradient of the outputs predicted by the model (i.e. gradient of y_pred). For example, the loss function may try to minimize the mean squared error between the gradient of the model output (grad(y_pred)) and y_true.

Although many operations like taking the square of y_pred inside the custom loss function is okay, using its gradient within the custom loss function is lnot possible. I understand that this might be because y_pred is simply a reference to a computational graph, and doesn't have any values initially. Can someone please suggest a workaround for this?

tensorflow keras loss-function

share|improve this question

asked Mar 9 at 3:44

Tarak Nath NandiTarak Nath Nandi

1

• 
  
  
  

  
  

  
  
  
  
  
  

  
  
  Possible duplicate of Make a custom loss function in keras
  
  – alift
  Mar 9 at 6:40
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  
  It's not a duplicate of the question your link refers to. This is not about how to create a custom loss function, this is about doing a specific operation within it.
  
  – Tarak Nath Nandi
  Mar 9 at 15:37
  

  
  
  
  

add a comment | 

0

I'm trying to build a custom loss function which utilizes the gradient of the outputs predicted by the model (i.e. gradient of y_pred). For example, the loss function may try to minimize the mean squared error between the gradient of the model output (grad(y_pred)) and y_true.

Although many operations like taking the square of y_pred inside the custom loss function is okay, using its gradient within the custom loss function is lnot possible. I understand that this might be because y_pred is simply a reference to a computational graph, and doesn't have any values initially. Can someone please suggest a workaround for this?

tensorflow keras loss-function

share|improve this question

asked Mar 9 at 3:44

Tarak Nath NandiTarak Nath Nandi

1

• 
  
  
  

  
  

  
  
  
  
  
  

  
  
  Possible duplicate of Make a custom loss function in keras
  
  – alift
  Mar 9 at 6:40
  

  
  
  
  


• 
  
  
  

  
  

  
  
  
  
  
  

  
  
  It's not a duplicate of the question your link refers to. This is not about how to create a custom loss function, this is about doing a specific operation within it.
  
  – Tarak Nath Nandi
  Mar 9 at 15:37
  

  
  
  
  

add a comment | 

I'm trying to build a custom loss function which utilizes the gradient of the outputs predicted by the model (i.e. gradient of y_pred). For example, the loss function may try to minimize the mean squared error between the gradient of the model output (grad(y_pred)) and y_true.

Although many operations like taking the square of y_pred inside the custom loss function is okay, using its gradient within the custom loss function is lnot possible. I understand that this might be because y_pred is simply a reference to a computational graph, and doesn't have any values initially. Can someone please suggest a workaround for this?

tensorflow keras loss-function

share|improve this question

asked Mar 9 at 3:44

Tarak Nath NandiTarak Nath Nandi

1

share|improve this question

asked Mar 9 at 3:44

Tarak Nath NandiTarak Nath Nandi

1

share|improve this question

asked Mar 9 at 3:44

Tarak Nath NandiTarak Nath Nandi

1

asked Mar 9 at 3:44

Tarak Nath NandiTarak Nath Nandi

1

asked Mar 9 at 3:44

Tarak Nath NandiTarak Nath Nandi

1