[solved] – Question 95188

A Table 1 have two inputs (x1,x2), and 1 output d.
And we first train the first 20 training epochs progress under the certain conditions itemized for table 1.
Then change the representation of x1, x2 as input for Table 2.
Then, plotting the sum square error in the two Tables for each training epoch will show how effectively your perceptron converges (i.e. the training curve=learning curve).
.Finally, we can conclude that the learning curve of Table 1 converges better than the learning curve of Table 2.
question: Does here exist any impact on the perceptron’s performance ? if exist, what is the main reason when learning algorithm itself has not changed?

Expert Answer

---