lecture 15 done. Based on the insights I got from this lecture,I have these two questions, Q1: Since you said that we particularly don't have the information when the data lies within 0.5 even when we are using sigmoid function, does this mean it has it's own limitation and not the best optimization technique to use? Q2: Is the only use case of regularizer to prevent the ML models from overfitting issue?
For question 1 , I think sir here use sigmoid in place of signum function because it have discontinuity as x=0, its out put directly 1,0 . But in case of sigmoid it follows the exponential so . We got for information about points
Sigmoid_visual_3D is still not working on my computer - the following error message appears - NameError: name 'boundary_x' is not defined Luis Carlos Timm from Brazil
lecture 15 done. Based on the insights I got from this lecture,I have these two questions, Q1: Since you said that we particularly don't have the information when the data lies within 0.5 even when we are using sigmoid function, does this mean it has it's own limitation and not the best optimization technique to use? Q2: Is the only use case of regularizer to prevent the ML models from overfitting issue?
For question 1 , I think sir here use sigmoid in place of signum function because it have discontinuity as x=0, its out put directly 1,0 . But in case of sigmoid it follows the exponential so . We got for information about points
For question 2 , I think
To avoid over fitting but I also don’t know how it avoid over fitting
Sigmoid_visual_3D is still not working on my computer
- the following error message appears - NameError: name 'boundary_x' is not defined
Luis Carlos Timm from Brazil
Sigmoid_visual
_3d.ipynb fileis is not working.. looks like file is corrupted..
You are right. It has been replaced now. Please check the modified file