An AI model developed for facial recognition is found to have significantly lower accuracy for certain ethnic groups. How would you approach correcting this bias without compromising the model’s overall accuracy?
- Remove support for the affected ethnic groups.
- Fine-tune the model using additional data from the underrepresented groups.
- Ignore the issue as it's impossible to fix.
- Rerun the model on the same data to validate the bias.
To correct bias in facial recognition AI, it's crucial to fine-tune the model using additional data from the underrepresented ethnic groups. This helps improve accuracy without compromising fairness.
Loading...
Related Quiz
- What is the primary purpose of using robo-advisors in financial planning?
- The concept of _______ involves machines being able to learn from data without being explicitly programmed.
- The learning rate in an optimization algorithm affects how _______ the model converges to a minimum.
- _______ refers to a decentralized approach to machine learning where the model is trained across multiple decentralized devices.
- In the context of transfer learning, what is a potential risk when adapting a pre-trained model to a new task?