Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Sigmoid for multi-class classification?

See original GitHub issue

@qubvel I saw on the readme you mentioned using a sigmoid activation function for a multi-class classification problem:

# multiclass segmentation with independent overlapping/non-overlapping class masks
model = sm.Unet('resnet34', classes=3, activation='sigmoid')

What’s the intuition for this and do you have any resources you can point me to? I’ve never thought to do this but for some reason with my dataset it actually does slightly better?

thanks

Issue Analytics

State:
Created 4 years ago
Reactions:1
Comments:5 (1 by maintainers)

Top GitHub Comments

1reaction

junxnonecommented, Mar 24, 2020

thanks，it’s very clear

1reaction

JordanMakesMapscommented, Nov 13, 2019

Ah, sorry, this is an instance for multi-label scenarios? That makes sense as to why this works slightly better with my dataset. Thanks!

Top Results From Across the Web

How does Sigmoid activation work in multi-class classification ...

If your task is a kind of classification that the labels are mutually exclusive, each input just has one label, you have to...

Multi-label vs. Multi-class Classification: Sigmoid vs. Softmax

When you have multiple reasonable classifier outputs, use a “moid” (sigmoid – the two “moids”/”maids” on the left of the picture). When you...

Can I use a sigmoid function with multiclass classification?

Yes you can, but i recommend that you use sigmoid when your data can belong to more then 1 class at a time....

Sigmoid activation for multi-class classification? - Stack Overflow

Sigmoids are activation functions of the form 1/(1+exp(-z)) where z is the scalar multiplication of the previous hidden layer (or inputs) and a ......

Sigmoid and SoftMax Functions in 5 minutes

Sigmoid is used for binary classification methods where we only have 2 classes, while SoftMax applies to multiclass problems. In fact, the SoftMax...