Câu 25: RAI: Risk and Artificial Intelligence

A company is developing a model to classify network traffic as malicious or benign. The data includes both labeled and unlabeled datasets. A data scientist emphasizes that the success of any model using this data highly depends on certain structural properties of the data. Which of the following best illustrates conditions in which the model is likely to perform well?

1. A. Malicious and benign traffic observations form distinct clusters and the boundary between them lies in a low density region. — đáp án hiện tại
2. B. The unlabeled and labeled data have different features to prevent overfitting the model.
3. C. Labeled observations are scattered randomly and the boundary separating the two classes lies in a dense region of the data.
4. D. The unlabeled and labeled data have entirely different distributions, with no overlapping data patterns across the two groups.