Aporia Raised $25M Series A to Build Trust in AI 🎉 Read More
Aporia Raised $25M Series A 🎉 Read More


Post-Processing is defined as the processing of a model’s output after the model has been run. This is useful for enforcing fairness constraints without necessarily modifying the model. 


For instance, one might post-process a binary classifier by setting a classification threshold to ensure that equal opportunity for a given attribute is maintained by checking whether true positive rates are the same regardless of the attribute in question.