Neural Networks

Question 1

Beschreib Hebbs Rule

Answer 1

Question 2

Beschreib den Aufbau eines Perceptrons

Answer 2

Question 3

Wie lautet die Funktion der inner activation?

Answer 3

Question 4

Wie funktioniert das Aktualisieren der Gewichte?

Answer 4

Question 5

Wie minimiert man E?

Answer 5

Über das Gradientenabstiegsverfahren

Question 6

Beschreib die Rule for the change of weights mathematisch

Answer 6

Question 7

Wie funktioniert der Perceptron Algorithmus?

Answer 7

Question 8

Wie funktionert lineare Separierbarkeit mir Perceptrons?

Answer 8

Question 9

Was sind Sigmoid Neurons?

Answer 9

Haben folgende Aktivierungsfunktion

Question 10

Why not choose all-same input weights for a layer?

Answer 10

all units will have same activations → information loss

Question 11

Why not choose all-same output weights for a layer?

Answer 11

learning signals will be same → symmetry-breaking problem

Question 12

Wie funktioniert Error Backpropagation?

Answer 12

Question 13

Leite die Sigmoid Activation Function ab

Answer 13

Question 14

Beschreib Batch update

Answer 14

Question 15

Beschreib online update

Answer 15

Question 16

Beschreib Minibatch Updates

Answer 16

Question 17

Was ist der Momentum Term? Was sind die Vorteile davon?

Answer 17

Question 18

Which of the statements about convolutional neural networks (CNNs) are true?
1. With CNNs, one can put emphasis on certain objects and/or areas in an image to increase classification accuracy.
2. The kernel size has to be chosen greater than the size of the input image.
3. CNNs can capture dependencies between pixels, which distinguishes them from “classic” feed-forward neural networks.
4. The filters used in CNNs are manually designed for each dataset.

Answer 18

1 und 3

Question 19

Which of the following statements about the Perceptron algorithm are true?
1. At the start of training a perceptron, all weigths are initialized with 0.
2. Before updating the weights, the errors for all datapoints seen so far have to be calculated.
3. The value of the cost function depends both on the weight parameters of the perceptron and the input.
4. The weights are updated by the gradient of the cost function scaled by the learning rate.

Answer 19

3 und 4

Question 20

Which of the following statements about Receiver Operating Characteristic (ROC) curves is true?
1. ROC curves are used to measure the harmonic mean of precision and recall.
2. ROC curves are used to visualize the relationship between sensitivity and specificity.
3. ROC curves plot a function dependent on the False Positive rate.
4. The AUC (Area Under Curve) can be used as a measure of model performance.

Answer 20

2 und 4

Question 21

Which of the functions below are commonly used as activation functions in deep neural networks?
1. Hyperbolic Tangent (tanh)
2. Hyperbolic Secant (sech)
3. Softmax
4. Rectified Linear Unit (ReLU)
5. Sigmoid

Answer 21

alles außer 2

Question 22

One major disadvantage of linear activation functions is that they often suffer from vanishing gradients, i.e. weight adaptation saturates too quickly during the training process. Stimmt das?

Answer 22

Falsch

Question 23

Beschreib die Aktivierungsfunktion Hyperbolic Tangent (tanh)

Answer 23

Question 24

Was sind Vorteile von tanh?

Answer 24

• Outputs are zero-centered, which can help with faster convergence in training
• Stronger gradients compared to the sigmoid function

Question 25

Was sind Nachteile von tanh?

Answer 25

Can still suffer from the vanishing gradient problem for very large or small inputs

Question 26

Beschreib Softmax als Aktivierungsfunktion

Answer 26

Question 27

Beschreib die Aktivierungsfunktion Rectified Linear Unit (ReLU)

Answer 27

Question 28

Was sind Vorteile von ReLU?

Answer 28

• Computationally efficient
• Helps mitigate the vanishing gradient problem
• Allows for sparse activation in neural networks

Question 29

Was sind Nachteile von ReLU?

Answer 29

Can suffer from the “dying ReLU” problem, where neurons can become inactive and stop learning

Question 30

Was sind Vorteile der Sigmoid Funktion?

Answer 30

• Outputs can be interpreted as probabilities
• Smooth gradient

Question 31

Was sind Nachteile der Sigmoid Funktion?

Answer 31

• Outputs are not zero-centered
• Can suffer from vanishing gradient problem for very large or small inputs