Neural Networks and GANs Quiz

The tanh activation usually works better than sigmoid activation function for hidden units because the mean of its output is closer to zero, and so it centers the data better for the next layer. True/False?

• True

• False

You are building a binary classifier for recognizing cucumbers (y=1) vs. watermelons (y=0). Which one of these activation functions would you recommend using for the output layer?

• ReLU

• Leaky ReLU

• sigmoid

• tanh

The weights of a neural network should be initialized randomly rather than to all zeros, because if you initialize to all zeros, then the network will fail to break symmetry and will not converge, True/False?

• True

• False

Derivative of sigmoid activation function is: f(x) = 1/(1+e-x)

• f(x) (1-f(x))

• f(x) (1+f(x))

• f(x) ln(1-f(x))

• f(x) ln(1+f(x))

A binary classifier predicts "Positive" for 80 samples. Out of those, 60 are actually Positive. What is the number of True Positives (TP)?

• 60

• 20

• 80

• 40

A classifier predicted 100 negatives, out of which 90 are actually negative. What is the number of True Negatives (TN)?

• 10

• 90

• 100

• 110

What is true about the F1-score?

• It is geometric mean of precision and recall

• It is arithmatic mean of precision and recall

• It is harmonic mean of precision and recall

• The relationship can't be established

Which metric evaluates how well a model distinguishes between genuine and imposter pairs?

• Accuracy

• Confusion Matrix

• ROC Curve (Receiver Operating Characteristic)

• MAE (Mean Absolute Error)

What does GAN stand for in machine learning?

• Generative Autonomous Network

• Generative Adversarial Network

• Generalized Artificial Neuron

• General Adversarial Network

Who proposed GANs in 2014?

• Ian Goodfellow

• Geoffrey Hinton

• Yann LeCun

• Yoshua Bengio