search

shivashankarrs / classimb_fairness

5 stars 0 forks source link

Task 6 (design experiments) #6

Open shivashankarrs opened 3 years ago

shivashankarrs commented 3 years ago

--- for deepmoji, vary proportion of positive/negative, and the proportion of protected attributes as well (2 variables) --- for biography, lets keep the data as it is, the class imbalance and gender imbalance in professions, we can just handle whatever is there in the dataset.

shivashankarrs commented 3 years ago

So, we need a simulator for deepmoji alone.