Highly dependable and predictable, easily interpretable and modifiable, inexpensive

Highly dependable and predictable, easily interpretable and modifiable, inexpensive

• A. Advantages of rule-based approach
• B. Disadvantages of machine learning approaches
• C. Transformer architecture
• D. Generative adversarial networks (GANs)

Answer: (A)

The main idea is that rule-based systems can be highly dependable, predictable, interpretable, easily modifiable, and inexpensive. These systems rely on explicit if-then rules defined by domain experts. Because decisions follow those fixed rules, outputs are deterministic for a given input, making the behavior dependable and easy to forecast. You can trace every decision back to a specific rule, which makes the logic straightforward to understand and audit. Modifications are simply updating or adding rules, which avoids retraining a model and keeps maintenance inexpensive. In contrast, many machine learning approaches learn from data and can be less predictable, requiring large datasets, substantial training, and ongoing computational resources, which increases cost and can reduce interpretability. Advanced ML architectures like transformers or GANs are powerful but tend to be resource-intensive and harder to explain, so they don’t fit as neatly with being easily interpretable and inexpensive.

The main idea is that rule-based systems can be highly dependable, predictable, interpretable, easily modifiable, and inexpensive. These systems rely on explicit if-then rules defined by domain experts. Because decisions follow those fixed rules, outputs are deterministic for a given input, making the behavior dependable and easy to forecast. You can trace every decision back to a specific rule, which makes the logic straightforward to understand and audit. Modifications are simply updating or adding rules, which avoids retraining a model and keeps maintenance inexpensive. In contrast, many machine learning approaches learn from data and can be less predictable, requiring large datasets, substantial training, and ongoing computational resources, which increases cost and can reduce interpretability. Advanced ML architectures like transformers or GANs are powerful but tend to be resource-intensive and harder to explain, so they don’t fit as neatly with being easily interpretable and inexpensive.