How’s Softmax and Sigmoid Classification Different?

There are a lot of deep explanations elsewhere so here I’d like to share tips on what you can say during an interview setting.

In a classification model for 3 labels (e.g. classifying a teddy bear, a grizzly bear and a black bear in image recognition), can a softmax multi-class classification network still work, when there are cases where more than one target appears (e.g. both teddy bear and grizzly bear appear in the same image)?

Here are some example answers for readers’ reference:

The simple answer is No. Softmax is used with the constraint that each example can only have one targeted label.

The following is an example illustration of how softmax and sigmoid differs.

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

Watch the explanation by Dr. Kian Katanforoosh from Stanford:

Check the explanation!

Happy practicing!