wayground logo

Free Printable Worksheets

Font size

S
M
L
XL
Worksheets

Artificial Neural Networks Worksheet Questions (Grade 13)

Total questions: 50

Worksheet time: 25mins

Name
Class
Date
1.

Which of the following is a characteristic of Artificial Neural Networks (ANN)?

a)

Linear processing only

b)

Distributed and parallel processing

c)

Rule-based computation

d)

Symbolic reasoning

2.

What is the primary function of dendrites in a biological neuron?

a)

Send signals

b)

Receive input signals

c)

Store information

d)

Generate impulses

3.

An artificial neuron is defined as:

a)

A biological cell

b)

A mathematical model of a neuron

c)

A rule-based system

d)

A digital switch

4.

Which activation function is typically used in multiclass classification?

a)

Sigmoid

b)

ReLU

c)

Tanh

d)

Softmax

5.

A continuous perceptron network uses:

a)

Binary outputs

b)

Continuous activation functions

c)

Step function only

d)

No weights

6.

In a Hopfield network, the weights are:

a)

Random

b)

Asymmetric

c)

Symmetric

d)

Always zero

7.

One learning difficulty in multilayer feedforward networks is:

a)

Overfitting

b)

Low storage

c)

High learning rate stability

d)

Zero training time

8.

The main objective of the backpropagation algorithm is to:

a)

Increase loss

b)

Minimize error by updating weights

c)

Increase the number of neurons

d)

Initialize weights randomly

9.

Data normalization techniques include:

a)

Min–Max scaling and Z-score normalization

b)

Fourier transform

c)

Gradient scaling

d)

Batch clipping

10.

Deep Belief Networks (DBN) use which learning method?

a)

Reinforcement learning

b)

Unsupervised pretraining

c)

Genetic learning

d)

Hebbian learning

11.

A biological neuron consists of:

a)

Dendrites, axon, soma

b)

Wires, sensors, chips

c)

Weights, bias, activation

d)

Inputs, outputs, loss

12.

A simple ANN architecture includes:

a)

Input, hidden, output layers

b)

Processor and memory

c)

Encoder and decoder

d)

Sensor and actuator

13.

A limitation of reinforcement learning is:

a)

Requires labeled data

b)

No interaction with environment

c)

No rewards

d)

Easy convergence

14.

A single-layer feedforward neural network has:

a)

One input and one output layer

b)

Multiple hidden layers

c)

Feedback connections

d)

Recurrent units

15.

If the learning rate of a perceptron is too large:

a)

Learning becomes unstable

b)

Convergence is faster

c)

The weights freeze

d)

Accuracy improves

16.

The generalized delta rule is used in:

a)

Perceptron learning

b)

Backpropagation learning

c)

Reinforcement learning

d)

Genetic algorithms

17.

Backpropagation uses which type of learning strategy?

a)

Supervised learning

b)

Unsupervised learning

c)

Semi-supervised learning

d)

Reinforcement learning

18.

Network normalization methods include:

a)

Batch normalization, Layer normalization

b)

Dropout, Early stopping

c)

Regularization only

d)

Max pooling

19.

Artificial vs Biological Neural Networks differ because:

a)

ANNs are digital; biological are analog

b)

Both process chemically

c)

Both are hardware-based

d)

ANNs use DNA-based memory

20.

Total parameters in a fully connected network with 3 input, 5 hidden, and 10 output neurons:

a)

15

b)

80

c)

40

d)

60

21.

Steps in backpropagation include:

a)

Forward pass → Error calculation → Backward weight update

b)

Weight freeze → Output reset

c)

Only error calculation

d)

Random updates

22.

Deep feedforward neural networks contain:

a)

Multiple hidden layers

b)

Only one output neuron

c)

Recurrent loops

d)

No bias terms

23.

Biological vs Artificial Neuron difference:

a)

Biological uses electrochemical signals; Artificial uses mathematical functions

b)

Both are identical

c)

Artificial uses chemical signals

d)

Biological uses digital pulses

24.

Hopfield network energy function minimizes:

a)

System error

b)

Output variance

c)

Entropy

d)

Gradient

25.

Purpose of backpropagation:

a)

Update weights to reduce error

b)

Freeze weights

c)

Increase bias

d)

Skip training

26.

Machine Learning vs Deep Learning difference:

a)

Deep learning uses neural networks with multiple layers

b)

Machine learning uses deep layers

c)

Deep learning uses rules

d)

ML does not learn

27.

Structure of Deep Belief Network includes:

a)

Stacked Restricted Boltzmann Machines

b)

Feedforward perceptrons

c)

CNN layers

d)

RNN cells

28.

The sigmoid function output range is:

a)

(-1, 1)

b)

(0, 1)

c)

(-∞, ∞)

d)

(0, ∞)

29.

The tanh activation function output range is:

a)

(0, 1)

b)

(-1, 1)

c)

(-∞, ∞)

d)

(0, ∞)

30.

ReLU stands for:

a)

Rectified Linear Unit

b)

Relative Linear Unit

c)

Reduced Logic Unit

d)

Recursive Layer Unit

31.

Gradient vanishing occurs in:

a)

Deep networks with sigmoid/tanh

b)

Shallow networks only

c)

CNNs always

d)

RNNs with ReLU

32.

In supervised learning, training data contains:

a)

Only inputs

b)

Inputs with labeled outputs

c)

Only outputs

d)

Random signals

33.

Dropout is used to:

a)

Prevent overfitting

b)

Reduce training data

c)

Increase bias

d)

Increase learning rate

34.

Batch normalization helps to:

a)

Stabilize and speed up training

b)

Reduce batch size

c)

Eliminate gradients

d)

Decrease hidden units

35.

Learning rate controls:

a)

Step size in weight updates

b)

Number of neurons

c)

Network depth

d)

Output value

36.

Hebbian learning principle states:

a)

Neurons that fire together wire together

b)

Neurons that rest together stay together

c)

Random firing

d)

Decrease in weights

37.

The main objective of normalization:

a)

Scale inputs for faster convergence

b)

Change feature relationships

c)

Randomize data

d)

Reduce sample size

38.

The bias in a neuron helps to:

a)

Shift the activation function

b)

Scale inputs

c)

Increase learning rate

d)

Reduce weights

39.

Which network uses energy minimization principle?

a)

Hopfield Network

b)

CNN

c)

RNN

d)

Perceptron

40.

The softmax function outputs:

a)

Probabilities that sum to one

b)

Binary outputs

c)

Negative values

d)

Random scores

41.

Epoch in neural network training refers to:

a)

One complete pass through training data

b)

A single batch update

c)

A test cycle

d)

Random initialization

42.

Overfitting occurs when:

a)

Model learns noise in training data

b)

Model underperforms on training data

c)

Model generalizes well

d)

Learning rate is small

43.

Which of the following is a deep learning framework?

a)

TensorFlow

b)

MATLAB

c)

Excel

d)

Power BI

44.

Reinforcement learning learns by:

a)

Reward and punishment signals

b)

Predefined labels

c)

Random guessing

d)

Manual updates

45.

In a discrete Hopfield network, the weight matrix is typically...

a)

Asymmetric

b)

Symmetric (Wij=Wji)

c)

Trained using backpropagation

d)

A diagonal matrix

46.

Synaptic transmission, the communication between neurons, is primarily mediated by chemicals called...

a)

Neurotransmitters

b)

Enzymes

c)

Hormones

d)

Electrolytes

47.

What are the two main 'passes' of the backpropagation algorithm?

a)

Input Pass and Output Pass

b)

Normalization Pass and Training Pass

c)

Weight Initialization and Error Calculation

d)

Forward Pass and Backward Pass

48.

The Hopfield network is a form of...?

a)

Feedforward Neural Network

b)

Recurrent Neural Network

c)

Convolutional Neural Network

d)

Autoencoder

49.

What is the primary purpose of the 'activation function' in an artificial neuron?

a)

To sum the weighted inputs

b)

To normalize the input data

c)

To introduce non-linearity into the network

d)

To store the weight value

50.

A fully connected deep feedforward network is characterized by information flowing...

a)

Only in one direction, from input to output, through the layers

b)

In cycles, allowing for memory

c)

Only between neurons in the same layer

d)

From the output layer back to the input layer