Traditional machine learning and deep learning are the core technologies of artificial intelligence. There is a slight difference in the engineering process. The following steps. What you don't need…

Question

Options

• AModel evaluation
• BFeature engineering
• CData cleaning
• DModel building

• A
  8% (4)
• B
  74% (39)
• C
  4% (2)
• D
  15% (8)

Explanation

Feature engineering (B) is the step deep learning eliminates. Deep learning models - particularly neural networks - automatically learn relevant features directly from raw data through their layered architecture, removing the need for humans to manually craft and select input features.

Why the distractors are wrong:

• A (Model evaluation) is always required - you must measure accuracy, loss, and generalization regardless of the algorithm used.
• C (Data cleaning) is always required - garbage in, garbage out applies equally to deep learning; noisy or missing data still degrades performance.
• D (Model building) is obviously required - you still design and train the neural network architecture itself.

Memory tip: Think of deep learning as "feature-free engineering" - the deep layers do the feature extraction for you, so the human engineer steps back from that task. If you remember "deep = automatic features," option B will always stand out.

No community discussion yet for this question.