sse

(To be removed) Sum squared error performance function

sse will be removed in a future release. For more information, see Transition Legacy Neural Network Code to dlnetwork Workflows.

For advice on updating your code, see Version History.

Syntax

perf = sse(net,t,y,ew)

perf = sse(net,t,y,ew,Name,Value)

Description

perf = sse(net,t,y,ew) takes a network net, targets T, outputs Y, and optionally error weights EW, and returns network performance calculated as the sum squared error.

sse is a network performance function. It measures performance according to the sum of squared errors.

perf = sse(net,t,y,ew,Name,Value) has two optional function parameters that set the regularization of the errors and the normalizations of the outputs and targets.

sse is a network performance function. It measures performance according to the sum of squared errors.

Examples

Calculate Network Performance with 'sse' Function

This example shows how to calculate the performance of a feedforward network with the sse function.

Create a network using the data from the simple fit data set, train it, and calculate its performance.

[x,t] = simplefit_dataset;
net = fitnet(10);
net.performFcn = 'sse';
net = train(net,x,t)
y = net(x)
e = t-y
perf = sse(net,t,y)

Input Arguments

net — Input network

network

Input network, specified as a network object. To create a network object, use for example, feedforwardnet or narxnet.

t — Network targets

matrix | cell array

Network targets, specified as a matrix or cell array.

y — Network outputs

matrix | cell array

Network outputs, specified as a matrix or cell array.

ew — Network outputs

{1} (default) | vector | matrix | cell array

Error weights, specified as a vector, matrix, or cell array.

Error weights can be defined by sample, output element, time step, or network output:

ew = [1.0 0.5 0.7 0.2]; % Across 4 samples
ew = [0.1; 0.5; 1.0]; % Across 3 elements
ew = {0.1 0.2 0.3 0.5 1.0}; % Across 5 timesteps
ew = {1.0; 0.5}; % Across 2 outputs

The error weights can also be defined across any combination, such as across two time-series (i.e., two samples) over four timesteps.

ew = {[0.5 0.4],[0.3 0.5],[1.0 1.0],[0.7 0.5]};

In the general case, error weights may have exactly the same dimensions as targets, in which case each target value will have an associated error weight.

The default error weight treats all errors the same.

ew = {1}

Name-Value Arguments

Example: 'regularization',0.1

Specify optional comma-separated pairs of Name,Value arguments. Name is the argument name and Value is the corresponding value. Name must appear inside quotes. You can specify several name and value pair arguments in any order as Name1,Value1,...,NameN,ValueN.

regularization — Proportion of performance

0 (default) | integer between 0 and 1

Proportion of performance attributed to weight and bias values, specified as the comma-separated pair consisting of 'regularization' and an integer between 0 and 1. The larger this value is, the more the network is penalized for larger weights, and the more likely the network function avoids overfitting.

normalization — Output and target normalization

'none' (default) | 'standard' | 'percent'

Output and target normalization, specified as the comma-separated pair consisting of 'normalization' and either:

• 'none' — performs no normalization.

• 'standard' — normalizes outputs and targets to [-1, +1], and therefore normalizes errors to [-2, +2].

• 'percent' — normalizes outputs and targets to [-0.5, +0.5], and therefore normalizes errors to [-1, +1].

Output Arguments

perf — Network performance

scalar

Network performance calculated as the sum squared error, returned as a scalar.

More About

Network Use

To prepare a custom network to be trained with sse, set net.performFcn to 'sse'. This automatically sets net.performParam to the default function parameters.

Then calling train, adapt or perform will result in sse being used to calculate performance.

Version History

Introduced before R2006a

R2026a: sse will be removed

sse will be removed in a future release.

To learn more about using metrics and loss functions when you train neural networks, see these examples and topics:

• Deep Learning Metrics

• Custom Loss Functions

• Monitor Deep Learning Training Progress

• ROC Curve and Performance Metrics

For more information, see Transition Legacy Neural Network Code to dlnetwork Workflows.