Loss Functions¶
MSELoss¶
-
class
recoder.losses.MSELoss(confidence=0, reduction='elementwise_mean')[source]¶ Computes the weighted mean squared error loss.
The weight for an observation x:
\[w = 1 + confidence \times x\]and the loss is:
\[\ell(x, y) = w \cdot (y - x)^2\]Parameters: - confidence (float, optional) – the weighting of positive observations.
- reduction (string, optional) – Specifies the reduction to apply to the output: ‘none’ | ‘elementwise_mean’ | ‘sum’. ‘none’: no reduction will be applied, ‘elementwise_mean’: the sum of the output will be divided by the number of elements in the output, ‘sum’: the output will be summed. Default: ‘elementwise_mean’
MultinomialNLLLoss¶
-
class
recoder.losses.MultinomialNLLLoss(reduction='elementwise_mean')[source]¶ Computes the negative log-likelihood of the multinomial distribution.
\[\ell(x, y) = L = - y \cdot \log(softmax(x))\]Parameters: reduction (string, optional) – Specifies the reduction to apply to the output: ‘none’ | ‘elementwise_mean’ | ‘sum’. ‘none’: no reduction will be applied, ‘elementwise_mean’: the sum of the output will be divided by the number of elements in the output, ‘sum’: the output will be summed. Default: ‘elementwise_mean’