AIF-C01 Question 170
Select 3Your company is developing a machine learning model to predict energy consumption for smart buildings. As part of responsible AI practices, the leadership team wants to ensure the model selection process considers environmental sustainability. Which of the following actions align with responsible model selection in this context?
- A
Choose a model that requires less computational power while maintaining acceptable accuracy.
- B
Select the most complex deep learning model available to ensure the highest accuracy, regardless of resource usage.
- C
Evaluate the energy consumption of training and inference for different model architectures.
- D
Use pre-trained models or transfer learning to reduce training time and computational resources.
- E
Prioritize models with the highest number of parameters for better performance, even if they consume more energy.
Show answer and explanation
Correct answers: A, C, D
Explanation
Responsible AI practices for model selection involve considering the environmental impact of computational resources used during training and inference. This includes selecting efficient models, evaluating energy consumption, and leveraging techniques like transfer learning to reduce computational requirements. These practices ensure sustainability while maintaining acceptable model performance.
- A. Correct.
This is correct because selecting a model that balances computational efficiency and accuracy is a key part of sustainable model selection.
- B. Incorrect.
This is incorrect because prioritizing complexity and resource usage without regard to environmental impact goes against responsible AI practices.
- C. Correct.
This is correct because evaluating energy consumption during model training and inference is a responsible practice to ensure environmental sustainability.
- D. Correct.
This is correct because using pre-trained models or transfer learning reduces the computational resources required, contributing to sustainability.
- E. Incorrect.
This is incorrect because prioritizing models with more parameters without considering resource usage or sustainability is not a responsible approach.