Smooth Loss Gradient Gradient for the smooth loss function

Smooth Loss Gradient Gradient for the smooth loss function

Usage

loss_smooth_grad(
  alpha,
  X,
  Y,
  epsilon,
  beta,
  lambda1 = 0,
  lambda2 = 0,
  weight = NULL
)

Arguments

alpha

The vector to calculate loss-gradient for

X

The data matrix

Y

The response vector

epsilon

The acceptable error

beta

The steepness of the sigmoid

lambda1

The L1 regulariser (default: 0)

lambda2

The L2 regulariser (default: 0)

weight

weight vector (default: NULL)

Value

The gradients for alpha