Exercise $5$

Instructions: You will maked a masked version of the accuracy function. You will need to perform an argmax to get the predicted label for each element in the batch. Remember to provide the appropriate axis in the argmax function. Furthermore, remember to use only tensorflow operations. Even though numpy has every function you will need, to pass it as a loss function and$/$or metric function, you must use tensorflow operations, due to internal optimizations that Tensorflow performs for reliable fitting. The following tensorflow functions are already loaded in memory, so you can directly call them.

tf$.$equal, equivalent to np$.$equal

tf$.$cast, equivalent to np$.$astype

tf$.$reduce$\_$sum, equiavalent to np$.$sum

tf$.$math.argmax, equivalent to np$.$argmax

You may need t# GRADED FUNCTION: masked$\_$accuracy

def masked$\_$accuracy$($y$\_$true, y$\_$pred$)$:

$"""$

Calculate masked accuracy for predicted labels.

Parameters:

y$\_$true $($tensor$)$: True labels.

y$\_$pred $($tensor$)$: Predicted logits$.$

Returns:

accuracy $($tensor$)$: Masked accuracy.

$"""$

### START CODE HERE ###

# Calculate the loss for each item in the batch.

# You must always cast the tensors to the same type in order to use them in training. Since you will make divisions, it is safe to use tf$.$float$32$ data type.

y$\_$true $=$ tf$.$cast$($y$\_$true, tf$.$float$32)$

# Create the mask, i$.$e$.,$ the values that will be ignored

mask $=$ None

mask $=$ tf$.$cast$($mask$,$ tf$.$float$32)$

# Perform argmax to get the predicted values

y$\_$pred$\_$class $=$ None

y$\_$pred$\_$class $=$ tf$.$cast$($y$\_$pred$\_$class, tf$.$float$32)$

# Compare the true values with the predicted ones

matches$\_$true$\_$pred $=$ tf$.$equal$($None$,$ None$)$

matches$\_$true$\_$pred $=$ tf$.$cast$($matches$\_$true$\_$pred $,$ tf$.$float$32)$

# Multiply the acc tensor with the masks

matches$\_$true$\_$pred $*=$ None

# Compute masked accuracy $($quotient between the total matches and the total valid values, i$.$e$.,$ the amount of non$-$masked values$)$

masked$\_$acc $=$ None$/$None

### END CODE HERE ###

return masktrue$\_$labels $=[0,1,2,0]$

predicted$\_$logits $=[[0\mathrm{.}1,0\mathrm{.}6,0\mathrm{.}3],[0\mathrm{.}2,0\mathrm{.}7,0\mathrm{.}1],[0\mathrm{.}1,0\mathrm{.}5,0\mathrm{.}4],[0\mathrm{.}4,0\mathrm{.}4,0\mathrm{.}2]]$

print$($masked$\_$accuracy$($true$\_$labels, predicted$\_$logits$))$ed$\_$accf.float$32$ while casting

Expected output:

tf$.$Tensor$(0\mathrm{.}5,$ shape$=(),$ dtype$=$float$32)$