Which claim regarding generative models isn't true?
Answer options
A
They can generate new data samples
B
They always require labeled data for training
C
They capture the data distribution
D
They can be combined with discriminative models for certain
E
tasks
F
They can be used in unsupervised learning scenarios
Correct answer: They always require labeled data for training
Explanation
The false claim is option [1]: 'They always require labeled data for training.' Generative models such as GANs, VAEs, and autoregressive language models can be trained in an unsupervised manner without requiring labeled data. All other options state true properties of generative models.