Hands-On Exercise: Implement a Simple Perceptron CT-AI Practice Questions (Page 1)

Questions 1–5

1. 1application · medium

   A team is training a perceptron on a linearly separable dataset. After training, the model correctly classifies all training samples. What can be concluded about the model's performance on unseen test data?

   • AThe model has overfit the training data and will perform poorly on test data.
   • BThe model's weights are guaranteed to be the optimal solution for the dataset.
   • CThe model will achieve 100% accuracy on unseen test data.
   • DThe model may not generalize perfectly to unseen data, and its test accuracy should be evaluated separately.
   Reveal Solution Discussion Select an answer first
2. 2expert · hard

   A QA team is testing a perceptron that classifies sensor readings as normal (0) or anomalous (1). The team wants to verify that the perceptron converges on a linearly separable dataset. Which test approach is most appropriate?

   • ATrain the perceptron and assert that the weights remain constant after the first epoch.
   • BTrain the perceptron on the dataset and assert that the number of misclassified samples reaches zero within a maximum number of epochs.
   • CTrain the perceptron and assert that the loss function decreases monotonically.
   • DTrain the perceptron and assert that the accuracy on the test set is 100%.
   Reveal Solution Discussion Select an answer first
4. 3application · medium

   A data scientist is using a perceptron to classify images as containing a cat (1) or not (0) based on two features: average color intensity and number of edges. The training data is not linearly separable. What is the expected outcome of training the perceptron?

   • AThe perceptron will converge to a solution that minimizes the number of errors.
   • BThe perceptron will not converge and will continue to make errors on some training samples.
   • CThe perceptron will converge to a solution that perfectly classifies all training samples.
   • DThe perceptron will fail to train because the activation function cannot handle this data.
   Reveal Solution Discussion Select an answer first
5. 4expert · hard

   A developer is implementing a perceptron and notices that the weights are updated even when the prediction is correct. What is the most likely cause of this bug?

   • AThe learning rate is set to 1, causing the weights to change on every sample.
   • BThe update rule is applied unconditionally, without checking if the prediction is correct.
   • CThe error is calculated as (predicted - true) instead of (true - predicted).
   • DThe activation function is not applied before computing the error.
   Reveal Solution Discussion Select an answer first
6. 5application · medium

   A developer is training a perceptron with learning rate η = 0.1. The current weights are [1.0, -0.5] and bias is 0.2. A training sample has features [2, 1] and true label 0. The perceptron predicts 1 (error = -1). What is the updated bias?

   • A0.0
   • B0.1
   • C0.2
   • D0.3
   Reveal Solution Discussion Select an answer first