# triplet_margin_with_distance_loss¶

paddle.nn.functional. triplet_margin_with_distance_loss ( input, positive, negative, distance_function=None, margin=1.0, swap=False, reduction='mean', name=None ) [source]

Measures the triplet loss given an input tensors $$x1$$, $$x2$$, $$x3$$ and a margin with a value greater than $$0$$. This is used for measuring a relative similarity between samples. A triplet is composed by input, positive and negative (i.e., input, positive examples and negative examples respectively). The shapes of all input tensors should be $$(N, D)$$.

The loss function for each sample in the mini-batch is:

$L(input, pos, neg) = \max \{d(input_i, pos_i) - d(input_i, neg_i) + {\rm margin}, 0\}$

where the default distance function

$d(x_i, y_i) = \left\lVert {\bf x}_i - {\bf y}_i \right\rVert_p$

or user can defined their own distance functions. margin is a nonnegative margin representing the minimum difference between the positive and negative distances that is required for the loss to be 0. If swap is true, it will compare distance of (input, negative) with distance of (negative, positive) and change it to the smaller one. For more details see http://www.bmva.org/bmvc/2016/papers/paper119/paper119.pdf.

Parameters
• input (Tensor) – Input tensor, the data type is float32 or float64. the shape is [N, *], N is batch size and * means any number of additional dimensions, available dtype is float32, float64.

• positive (Tensor) – Positive tensor, the data type is float32 or float64. The shape of label is the same as the shape of input.

• negative (Tensor) – Negative tensor, the data type is float32 or float64. The shape of label is the same as the shape of input.

• distance_function (callable, optional) –

Quantifies the distance between two tensors. if not specified, 2 norm functions will be used.

margin (float, optional):Default: $$1$$.A nonnegative margin representing the minimum difference between the positive and negative distances required for the loss to be 0.

• swap (bool, optional) – The distance swap changes the negative distance to the swap distance (distance between positive samples and negative samples) if swap distance smaller than negative distance. Default: False.

• reduction (str, optional) – Indicate how to average the loss by batch_size. the candicates are 'none' | 'mean' | 'sum'. If reduction is 'none', the unreduced loss is returned; If reduction is 'mean', the reduced mean loss is returned; If reduction is 'sum', the summed loss is returned. Default: 'mean'

• name (str, optional) – Name for the operation (optional, default is None). For more information, please refer to Name.

Returns

Tensor. The tensor variable storing the triplet_margin_with_distance_loss of input and positive and negative.

Return type

Output

Examples

import paddle

input = paddle.to_tensor([[1, 5, 3], [0, 3, 2], [1, 4, 1]], dtype=paddle.float32)
negative = paddle.to_tensor([[2, 1, -3], [1, 1, -1], [4, -2, 1]], dtype=paddle.float32)
loss = F.triplet_margin_with_distance_loss(input, positive, negative, margin=1.0, reduction='none')
print(loss)
# Tensor([0.        , 0.57496738, 0.        ])

loss = F.triplet_margin_with_distance_loss(input, positive, negative, margin=1.0, reduction='mean')
print(loss)
# Tensor([0.19165580])