Neural Network Exam Tests for Data Scientist

Exercise 1

What functions does the input layer of the multilayer perceptron perform?

1. Transmits a signal to the output layer of a multilayer perceptron.

2. Removes "noise" from the signal.

3. Transmits the input vector of signals to the first hidden layer.

4. Calculates the derivative for the backpropagation algorithm.

Task 2.

Is an axon a output or input process of a neuron?

1. Input

2. Output

Task 3.

What comes first - a mutation or a crossover? (in genetic algorithms).

1. Mutation

2. Crossover

3. Crossover operations in genetic algorithms do not exist.

4. Whatever.

Task 4.

Neuron j received at the input a signal from four other neurons excitation levels, the values ​​of which are 10, -20, 5, 4 and the corresponding weights of the connections are 0.8, 0.5, 0.7 and -0.5, respectively. Calculate the signal at the output of the j-th neuron in case the function of neuron activation is hyperbolic tangent ( ). Choose the correct answer:

one. -

2. 0.2449

3. 0.3145

4. 0.5

5. -0.5

Task 5.

Given: a neuron with an activation function of the type of hyperbolic tangent with three inputs. Inputs are all 1 and all weights are also 1. Parameter in the formula of the hyperbolic tangent is equal to 1. What will be the output of the neuron?

Option Y

Exercise 1.

What is usually longer - axon or dendrite?

1. Axon

2. Dendrite

Task 2.

Is it possible to use the “step” type activation function in the method of teaching backpropagation of an error?

1. Yes

2. No

Task 3.

Which of the following neural networks is a network with feedback?

1. Kohonen network.

2. Hamming network.

3. Grossberg output star.

4. Radial - basic network.

Task 4.

Neuron j received at the input a signal from four other neurons excitation levels, the values ​​of which are 10, -20, 5, 4 and the corresponding weights of the connections are 0.8, 0.5, 0.7 and -0.5, respectively. Calculate the signal at the output of the j-th neuron in case the function of neuron activation is a logistic sigmoid ( ). Choose the correct answer:

14

2. 0.25

3. -3.52

4. 1.7

Task 5.

Given: neural network with one hidden layer. The network has 1 input, 3 neurons in the hidden layer and one output. What will be the output of the network if the input 1, all the weight of the wound 1?

Exercise 1.

How can we determine in the Kohonen network which class the object belongs to?

1. Among the neurons of the Kohonen layer, choose the one with the maximum value.

2. Kohonen's network does not work with classification tasks.

3. The class number in the Kohonen network is determined by the combination of the outputs.

Task 2.

Dana Hamming's network. Let the learning examples be equal:

a) {1; 1; 1; 1} b) {1,1,1, -1} c) {1,1, -1, -1} and d) {1, -1, -1, -1 }

Write below what the weight matrix of the first layer will be equal to. :

Task 3.

Can after the initialization stage, the matrix of weights of synaptic weights of the Hopfield network look like this?

Task 4.

Can the number of memorized samples in the Hopfield network be less than the number of neurons?

1. Yes

2. No

Task 5.

What is the task of clustering objects using neural networks? Write below.

Option K

Exercise 1.

In the Hebbian differential method of learning, those neurons that _________

1. Inputs, which are least changed upwards

2. Outputs, which are most changed upwards

3. Synapses, which missed less information

Task 2.

Is it possible to apply letters to the inputs of the Kohonen network?

1. Yes

2. No

3. You can, but pre-encode their numbers

Task 3.

What is the Hamming distance between the words “Tanya” and “Katya”?

ten

2. 1

3. 2

4. 3

5. 4

Task 4.

A counterpropagation network is given, which has 6 neurons in the Kohonen layer and 8 neurons in the Grossberg layer. How many classes can the input examples be split using this network, if the Kohonen layer works in accreditation mode?

sixteen

2. 8

3

Task 5.

Does the Hopfield network work with signals equal to +1 and -1?

1. Yes

2. No