How to evaluate the performance by ANN

Hi,

I am working on ANN to recognize the characters. Here is the code. I want to know that how to evaluate the performance of trained network means accuracy of the network.

'trainingset' is the dataset for train the network and 'target' is shows the target vector of the trainingset.

close all; clear all;clc

load target;

%load train_1a;

% load trainingset

load trainingset

net = feedforwardnet([20 20], 'trainlm');

%net.performFcn = 'sse';

net.trainParam.lr=0.01;

net.trainParam.goal = 0.01;

net.trainParam.show = 20;

net.trainParam.epochs = 50;

%net.trainParam.mc = 0.95;

net.trainParam.min_grad =1e-15;

net.trainParam.max_fail =10;

net = train(net,trainingset,target);

Result = sim(net,trainingset) %sim it with dataset 'base'

Answer 1

[Jaimin]

Hi @neha gautam

MATLAB includes various tools and functions for assessing the performance of neural networks, such as calculating accuracy. The Deep Learning Toolbox (previously known as the Neural Network Toolbox) provides built-in functions specifically designed for performance evaluation.

A useful tool is the “confusion” function, which calculates the confusion matrix and accuracy for classification tasks.

To learn more about “confusion” function kindly refer following MathWorks Documentation.

https://www.mathworks.com/help/deeplearning/ref/confusion.html

I hope this will be helpful.