torch.js has not been released yet.

torch.nn.LPPool3d

class LPPool3d extends Module

new LPPool3d(norm_type: number, kernel_size: number | [number, number, number], options?: LPPool3dOptions)

readonlynorm_type(number)
readonlykernel_size(number | [number, number, number])
readonlystride(number | [number, number, number])
readonlyceil_mode(boolean)

3D Lp-norm pooling: reduces volumetric dimensions using Lp-norm.

Applies 3D spatial pooling using Lp-norm for robust feature extraction. Generalized pooling providing middle ground between max and average.

L1: Sum of absolute values
L2: Euclidean norm

3D robust pooling: Balance between max sensitivity and average smoothness
Memory intensive: 3D operations use more resources
Computational cost: Expensive but provides robust features

Examples

// 3D L2-norm pooling for volumetric data
const pool = new torch.nn.LPPool3d(2, [2, 3, 3]);  // L2-norm, mixed kernel
const volume = torch.randn([4, 64, 32, 64, 64]);
const y = pool.forward(volume);  // 3D Euclidean norm pooling

See Also

PyTorch torch.nn.LPPool3d

LPPool2dOptions

LPPool3dOptions