Understanding Bias in Machine Learning

What is a potential risk when a facial recognition system is trained using only images of women?

• It will recognize all genders equally.

• It may not accurately recognize men.

• It will have no impact on accuracy.

• It will be faster in processing images.

In machine learning, what does bias typically refer to?

• A model treating groups differently in an unfair way.

• The accuracy of predictions.

• The speed of data processing.

• A model's ability to learn quickly.

At what stage can bias enter the development of a machine learning model?

• Only during model deployment.

• At any point during development.

• Only during model evaluation.

• Only during data collection.

Why is it important to evaluate a model for bias?

• To ensure it runs faster.

• To detect bias from earlier stages.

• To make it more complex.

• To increase the model's size.

What is a consequence of training a heart disease model only with data from men?

• It will diagnose all diseases.

• It may not accurately diagnose women.

• It will be equally accurate for women.

• It will be faster in processing data.

What is an example of bias in data?

• Data that is too small.

• Data that is too large.

• Data that is not representative of all situations.

• Data that is too complex.

How can societal biases affect machine learning models?

• They make the model faster.

• They can be reflected in the model's training data.

• They have no effect.

• They simplify the model.

Why is it crucial to evaluate models for all groups?

• To increase their size.

• To ensure they contain as little bias as possible.

• To make them more complex.

• To make them run faster.

What can happen if a machine learning model is trained with biased data?

• It will be accurate for all situations.

• It can have a huge impact on society.

• It will be faster.

• It will be more complex.

What is the benefit of being aware of bias in machine learning?

• It increases the model's size.

• It helps to avoid bias and benefits more people.

• It makes the model run faster.

• It simplifies the model.