The squared error cost function (SECF) in logistic regression gives a non-convex graph, which leads to multiple local minima. Can I get proof of this non-convexity? I got an article on the non-convexity of the SECF of logistic regression by googling, and the screenshot is below. Here, one training …

Nonconvexity -logistic

paulinpaloalto November 5, 2023, 5:27pm 4

And here’s a thread locally that shows a graph of what the MSE cost surface looks like for Logistic Regression.

So just doing a little searching of the forums also turns up some answers.

1 Like

Topic		Replies	Views
Visualizing Squared Error Cost function for Logistic regression in 2D Supervised ML: Regression and Classification week-3	3	591	February 17, 2024
Why is Squared Error Cost for Logistic Regression non-convex? Supervised ML: Regression and Classification week-3	1	605	July 31, 2022
Non-convex graph is not the reason we do not use the "Error Cost Function" as we defined in linear regression Supervised ML: Regression and Classification week-3	5	598	June 19, 2022
Gradient descent C1_W1 Supervised ML: Regression and Classification week-1	2	514	July 31, 2022
Gradient Descent local max Supervised ML: Regression and Classification week-1	5	524	May 11, 2023