Activation

Abs

class paddle.v2.activation.Abs

Abs Activation.

Forward: \(f(z) = abs(z)\)

Derivative:

\[\begin{split}1 &\quad if \quad z > 0 \\ -1 &\quad if \quad z < 0 \\ 0 &\quad if \quad z = 0\end{split}\]

Exp

class paddle.v2.activation.Exp

Exponential Activation.

\[f(z) = e^z.\]

Identity

paddle.v2.activation.Identity

alias of Linear

Linear

class paddle.v2.activation.Linear

Identity Activation.

Just do nothing for output both forward/backward.

Log

class paddle.v2.activation.Log

Logarithm Activation.

\[f(z) = log(z)\]

Square

class paddle.v2.activation.Square

Square Activation.

\[f(z) = z^2.\]

Sigmoid

class paddle.v2.activation.Sigmoid

Sigmoid activation.

\[f(z) = \frac{1}{1+exp(-z)}\]

Softmax

class paddle.v2.activation.Softmax

Softmax activation for simple input

\[P(y=j|x) = \frac{e^{x_j}} {\sum^K_{k=1} e^{x_k} }\]

SequenceSoftmax

class paddle.v2.activation.SequenceSoftmax

Softmax activation for one sequence. The dimension of input feature must be 1 and a sequence.

result = softmax(for each_feature_vector[0] in input_feature)
for i, each_time_step_output in enumerate(output):
    each_time_step_output = result[i]

Relu

class paddle.v2.activation.Relu

Relu activation.

forward. \(y = max(0, z)\)

derivative:

\[\begin{split}1 &\quad if z > 0 \\ 0 &\quad\mathrm{otherwize}\end{split}\]

BRelu

class paddle.v2.activation.BRelu

BRelu Activation.

forward. \(y = min(24, max(0, z))\)

derivative:

\[\begin{split}1 &\quad if 0 < z < 24 \\ 0 &\quad \mathrm{otherwise}\end{split}\]

SoftRelu

class paddle.v2.activation.SoftRelu

SoftRelu Activation.

Tanh

class paddle.v2.activation.Tanh

Tanh activation.

\[f(z)=tanh(z)=\frac{e^z-e^{-z}}{e^z+e^{-z}}\]

STanh

class paddle.v2.activation.STanh

Scaled Tanh Activation.

\[f(z) = 1.7159 * tanh(2/3*z)\]

SoftSign

class paddle.v2.activation.SoftSign

SoftSign Activation.

\[f(z)=\frac{z}{1 + |z|}\]