Objective 3.4 Identify and Mitigate Fairness and Non-Discrimination (bias) Risks CERTIFIED-ETHICAL-EMERGING-TECHNOLOGIST Practice Questions (Page 4)

Questions 16–20

1. 16expert · hard

   A data science team is using AI Fairness 360 (AIF360) to mitigate bias in a credit scoring model. They have identified that the model has a disparate impact on a certain racial group. They want to use a mitigation algorithm that can be applied to the trained model without retraining. Which AIF360 algorithm should they use?

   • AReweighing
   • BEqualized Odds Postprocessing
   • CAdversarial Debiasing
   • DDisparate Impact Remover
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A data science team is using AI Fairness 360 (AIF360) to mitigate bias in a hiring model. They have already identified that the model has a higher false-negative rate for female candidates. Which AIF360 module should they use to adjust the model's predictions to reduce this disparity?

   • AThe Fairness Metrics module to re-evaluate the model.
   • BThe Postprocessing module, such as Equalized Odds Postprocessing.
   • CThe Inprocessing module, such as Adversarial Debiasing.
   • DThe Preprocessing module, such as Reweighing.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A city government is deploying an AI system to predict which neighborhoods are at high risk of fire. The training data is based on historical fire incidents, which are more frequently reported in low-income neighborhoods due to underreporting in other areas. The model may therefore over-predict risk in low-income neighborhoods. Which type of bias is this an example of, and what is a suitable mitigation?

   • AAlgorithmic bias; apply a post-processing technique to adjust the model's risk scores.
   • BData bias; remove the low-income neighborhoods from the training data.
   • CHuman bias; conduct a bias audit to identify the source.
   • DData bias; use a data preprocessing technique to correct for underreporting, such as reweighting or synthetic data generation.
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   A data science team wants to use an open-source toolkit that provides a collection of fairness metrics and bias mitigation algorithms to assess and improve their models. Which tool is designed for this purpose?

   • ATensorFlow
   • BPandas
   • CAI Fairness 360
   • DKubernetes
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A financial services company is using Fairlearn to audit a credit risk model. They want to ensure that the model's error rates are similar across different age groups. Which Fairlearn metric should they use?

   • AIndividual fairness
   • BEqualized odds
   • CCalibration
   • DDemographic parity
   Reveal Solution Discussion Select an answer first