You are on page 1of 4

CSE&S Department BUET Khuzdar

Submitted by: Muhammad Younus


Roll No.: 21MS-CS22
Semester: 3rd
Subject: Machine learning
Submitted to: Engr.Noor Ahmed
Result with original code:

Changing softmax to sigmoid:


Changing tanh to relu:

Changing n_sambles 1000 to 500 and n_train 500 to 250:


Changing binary_crossentropy to mean_squared_error:

Summary:
After changing parameters and hyperparameters of the code at
google colab the result varies from changes.
I learn that different results on same changed values, this
means that the model learns over time after performing several
times and model is capable for further learning.

You might also like