sigmoidLayer

Sigmoid layer

Description

A sigmoid layer applies a sigmoid function to the input such that the output is bounded in the interval (0,1).

Tip

To use the sigmoid layer for binary or multilabel classification problems, create a custom binary cross-entropy loss output layer or use a custom training loop.

Creation

Syntax

layer = sigmoidLayer

layer = sigmoidLayer('Name',Name)

Description

layer = sigmoidLayer creates a sigmoid layer.

layer = sigmoidLayer('Name',Name) creates a sigmoid layer and sets the optional Name property using a name-value pair argument. For example, sigmoidLayer('Name','sig1') creates a sigmoid layer with the name 'sig1'. Enclose the property name in single quotes.

Properties

Name — Layer name

'' (default) | character vector | string scalar

Layer name, specified as a character vector or a string scalar. To include a layer in a layer graph, you must specify a nonempty unique layer name. If you train a series network with the layer and Name is set to '', then the software automatically assigns a name to the layer at training time.

Data Types: char | string

NumInputs — Number of inputs

1 (default)

Number of inputs of the layer. This layer accepts a single input only.

Data Types: double

InputNames — Input names

{'in'} (default)

Input names of the layer. This layer accepts a single input only.

Data Types: cell

NumOutputs — Number of outputs

1 (default)

Number of outputs of the layer. This layer has a single output only.

Data Types: double

OutputNames — Output names

{'out'} (default)

Output names of the layer. This layer has a single output only.

Data Types: cell

Examples

Create Sigmoid Layer

Create a sigmoid layer with the name 'sig1'.

layer = sigmoidLayer('Name', 'sig1')

layer = 
  SigmoidLayer with properties:

    Name: 'sig1'

  Show all properties

More About

Sigmoid Layer

A sigmoid layer applies a sigmoid function to the input such that the output is bounded in the interval (0,1).

This operation is equivalent to

$$f(x)=\frac{1}{1+e^{-x}}.$$

A multilabel classification problem can be thought of as a binary classification problem, where each class is considered independently of other classes as either present or not present. Solving this type of problem requires the sigmoid activation function, where for any sample x_n the posterior probability of class C_k is

$$\text{P(}{C}_k|x_n)=\frac{1}{1+e^{-a_k}}.$$

The value a_k is the weighted sum of all the units that are connected to class k. Performing multilabel classification requires a sigmoid layer followed by a custom binary cross-entropy loss layer.

Extended Capabilities

C/C++ Code Generation
Generate C and C++ code using MATLAB® Coder™.

GPU Code Generation
Generate CUDA® code for NVIDIA® GPUs using GPU Coder™.