mindspore.ops.multilabel_margin_loss

mindspore.ops.multilabel_margin_loss(input, target, reduction='mean')

Hinge loss for optimizing a multi-label classification.

Creates a criterion that optimizes a multi-label multi-classification hinge loss (margin-based loss) between input $x$ (a 2D mini-batch Tensor) and output $y$ (which is a 2D Tensor of target class indices). For each sample in the mini-batch:

$$\text{loss}(x,y)=\sum _{ij}\frac{max(0,1-(x[y[j]]-x[i]))}{\text{x.size}(0)}$$

where $x\in \{0,\cdots ,\text{x.size}(0)-1\}$, $y\in \{0,\cdots ,\text{y.size}(0)-1\}$, $0\le y[j]\le \text{x.size}(0)-1$, and $i\ne y[j]$ for all $i$ and $j$. $y$ and $x$ must have the same size. The criterion only considers a contiguous block of non-negative targets that starts at the front. This allows for different samples to have variable amounts of target classes.

Parameters

• input (Tensor) – Predict data, $x$ in the formula above. Tensor of shape $(C)$ or $(N,C)$, where $N$ is the batch size and $C$ is the number of classes. Data type must be float16 or float32.

• target (Tensor) – Ground truth data, $y$ in the formula above, with the same shape as input, data type must be int32 and label targets padded by -1.

• reduction (str, optional) –

  Apply specific reduction method to the output: 'none' , 'mean' , 'sum' . Default: 'mean' .

  • 'none': no reduction will be applied.

  • 'mean': compute and return the mean of elements in the output.

  • 'sum': the output elements will be summed.

Returns

• outputs (Union[Tensor, Scalar]) - The loss of MultilabelMarginLoss. If reduction is "none", its shape is $(N)$. Otherwise, a scalar value will be returned.

Raises

• TypeError – If input or target is not a Tensor.

• TypeError – If dtype of input is neither float16 nor float32.

• TypeError – If dtype of target is not int32.

• ValueError – If length of shape of input is neither 1 nor 2.

• ValueError – If shape of input is not the same as target.

• ValueError – If reduction is not one of 'none', 'mean', 'sum'.

Supported Platforms:

Ascend GPU

Examples

>>> import mindspore
>>> import numpy as np
>>> from mindspore import Tensor, ops
>>> inputs = Tensor(np.array([[0.1, 0.2, 0.4, 0.8], [0.2, 0.3, 0.5, 0.7]]), mindspore.float32)
>>> target = Tensor(np.array([[1, 2, 0, 3], [2, 3, -1, 1]]), mindspore.int32)
>>> output = ops.multilabel_margin_loss(inputs, target)
>>> print(output)
0.325