RAI Exam Preparation Material | Risk and AI (RAI)

Explanation:
Correct (C): Semi-structured data combines structured and unstructured elements, such as financial data with unstructured customer reviews.
Incorrect (A): Structured data is entirely organized in a set format without unstructured components.
Incorrect (B): Unstructured data lacks an organized format and does not include structured elements.
Incorrect (D): Binary data only represents attributes with two values and does not describe mixed data types.

Explanation:
B is correct. Early stopping prevents overfitting by halting training before the model perfectly fits the training data, which often generalizes better to test data
A is incorrect because increasing the learning rate may cause instability in training
C is incorrect as adding layers may exacerbate overfitting
D is incorrect as batch size does not directly address overfitting.

Explanation:
C is correct. The value function V(s) measures the expected cumulative future rewards, considering immediate and discounted future rewards, as described by the Bellman equation.
A is incorrect, as the policy function maps actions to states but doesn’t evaluate value
B is incorrect since it only provides immediate rewards without discounting future returns
D is incorrect as Q(s,a) calculates the action-value, not the state value.

Explanation:
C is correct. Transformers enable parallel processing, making them computationally efficient, especially for large text datasets
A is incorrect because although transformers use contextual embeddings, this does not relate directly to computational efficiency
B is incorrect as RNNs, not transformers, are sequence- dependent
D is incorrect as transformers can handle sequences without a strict order due to the attention mechanism.

Explanation:
Correct (B): When data points lie along the line y=x, it indicates the empirical and theoretical distributions are similar.
Incorrect (A): Scattered points suggest a poor fit.
Incorrect (C): Random distribution of points does not indicate a normal distribution match.
Incorrect (D): Concentration near the origin does not indicate fit and may imply outliers.