What people say about neural networks is wrong?

Question

Options

• AAs the number of hidden layers of the neural network increases, the classification ability of the
• BThe limitation of a single-layer perceptron is that it cannot solve the XOR problem
• CThe feedforward neural network can be represented by a directed acyclic graph
• DThere is no connection between neurons in the same layer of the feedforward neural network

• A
  73% (19)
• B
  4% (1)
• C
  8% (2)
• D
  15% (4)

Explanation

Option A is the incorrect statement because more hidden layers do not guarantee better classification ability. Beyond a certain depth, additional layers can hurt performance through problems like vanishing gradients, overfitting, and increased training difficulty - deeper is not always better without proper techniques like residual connections or careful regularization.

Options B, C, and D are all factually correct and therefore wrong answers to this question. The XOR limitation of single-layer perceptrons (B) is a classic, proven result because XOR is not linearly separable. Feedforward networks (C) genuinely are directed acyclic graphs since information flows in one direction with no cycles. And in standard feedforward architectures (D), neurons within the same layer truly have no lateral connections to one another.

Memory tip: When a question asks what is WRONG, look for the overstatement or absolute claim. "Increases always improve" language in neural network contexts is a red flag, since depth brings diminishing returns and new failure modes.

No community discussion yet for this question.