pmlayer.torch.layers.PMLinear

Linear layer with partially monotone constraints for PyTorch.

class PMLinear(
    num_input_dims,
    num_output_dims=1,
    indices_increasing=[],
    indices_decreasing=[],
    use_bias=True)

This layer applies a linear transformation to the incoming data \(y=Ax+b\). The output of this layer is monotonically increasing with respect to the input features specified by indices_increasing, and the output of this layer is monotonically decreasing with respect to the input features specified by indices_decreasing.

Parameters

Args	Type	Description
num_input_dims	int	The number of input features.
num_output_dims	int	The number of output features. Default: 1.
indices_increasing	list of int	The list of indices of monotonically increasing features. Default: [].
indices_decreasing	list of int	The list of indices of monotonically decreasing features. Default: [].
use_bias	bool	If set to False, the layer will not learn an additive bias. Default: True.

Tensor Shape

I/O	Shape
Input	(N, num_input_dims)
Output	(N, num_output_dims)

N usually represents the batch size.

Example

The following code transforms input tensor x into output tensor y, and the output (the shape of y) is (128,5).

l = PMLinear(10, 5, indices_increasing=[2,3])
x = torch.randn(128, 10)
y = l(x)
print(y.shape)