Multi Label Cross Entropy

• This is used in the output layer of multi label classification
• AKA Binary Cross Entropy
• Usually used after Sigmoid Function Layer
• For implementation: PyTorch Loss Functions

[!def] Multi Label Cross Entropy
$$
BCE = -( y log \hat{y} + (1-y) log (1- \hat{y})
$$
$$
\text{Total-CE Loss} = \frac{1}{M} \sum_i^M -( y_i log \hat{y_i} + (1-y_i) log (1- \hat{y_i})
$$