lolo icon indicating copy to clipboard operation
lolo copied to clipboard

Re-weight Gini impurity in multitask based on number of classes

Open maxhutch opened this issue 7 years ago • 0 comments

The maximum value of the Gini impurity is (n-1)/n, where n is the number of classes. This could cause multitask models to be biased towards modeling multi-class labels more accurately than binary classification ones.

maxhutch avatar Oct 09 '17 18:10 maxhutch