nntoolbox.components.maxout module

class nntoolbox.components.maxout.MaxoutLinear(in_features: int, out_features: int, nb_features: int, bias: bool = True)

Bases: torch.nn.modules.module.Module

A linear maxout layer:

output_i = max_{j = 1,…,k} (w_1 input + b_1, w_2 input + b_2,…, w_k input + b_k)

References:

Ian J. Goodfellow et al. “Maxout Networks.” https://arxiv.org/pdf/1302.4389.pdf

forward(input: torch.Tensor) → torch.Tensor

Parameters

input – (batch_size, in_features)

Returns

(batch_size, out_features)

training: bool