Quiz on nonlinear hypothesis, neurons and the brain, model representation, multiclass classification, cost function, gradient checking, and random initialization. Neural networks are a type of machine learning that uses algorithms to help computers learn and adapt without being reprogrammed. They are designed to mimic the human brain, with each neuron or node responsible for solving a small part of a problem. Neural networks can learn from the outputs they produce and the information they receive.  Neural networks use algorithms such as convolutional neural networks, recurrent neural... Show more

Quiz on nonlinear hypothesis, neurons and the brain, model representation, multiclass classification, cost function, gradient checking, and random initialization.

Neural networks are a type of machine learning that uses algorithms to help computers learn and adapt without being reprogrammed. They are designed to mimic the human brain, with each neuron or node responsible for solving a small part of a problem. Neural networks can learn from the outputs they produce and the information they receive. 

Neural networks use algorithms such as convolutional neural networks, recurrent neural networks, and deep neural networks. Each algorithm has unique differences that make it ideal for different applications. 

Here are some examples of neural networks:
Convolutional neural networks (CNNs): A deep learning algorithm that learns directly from data. CNNs are useful for finding patterns in images to recognize objects, classes, and categories. They can also be quite effective for classifying audio, time-series, and signal data.
Recurrent neural networks (RNNs): A deep learning model that is trained to process and convert a sequential data input into a specific sequential data output. RNNs are appropriate for applications where contextual dependencies are critical, such as time series prediction and natural language processing.
Feed forward neural networks: A basic type of neural network where information travels in only one direction from input to output. Feed forward neural networks are used in many applications, including machine translation, search engines, mobile applications, computer assistants, and object detection in photos. 
 

Related Test: Artificial Intelligence Practice Test: Neural Networks

Show less

Machine Learning 101 Practice Test: Neural Networks in Machine Learning

Time left 00:00

25 Questions

1. Which of the following statements is not true about the backpropagation?

Recurrent backpropagation is fed forward until a fixed value is achieved and after that, the error is computed and propagated backwardStatic backpropagation produces a mapping of a static input for static outputStatic backpropagation is useful to solve static classification issues like optical character recognitionThe mapping is static in both static back-propagation and recurrent backpropagation

2. Softmax layer is one of the best options as the output layer in a multiclass classification implementation.

Don't knowFalseMaybeTrue

3. How many hidden layers are present in the given figure?

1203

4. Why should one stop gradient checking once it is done before running the network for entire set of training iterations?

Because it would change the output of the training processBecause it would increase the speed of training processBecause it would slow down the speed of training processBecause it would nullify the output

5. The initialization and training of the weights on the interconnections through the training set require in the design of a neural network.

TrueDon't knowFalseMaybe

6. Which of the following statements is not true about neural networks?

NN has lots of interconnected nodes which are organized in layersThey are class of very powerful machine learning classifiersNeural networks are a class of computationally inexpensive algorithmsThese are capable of fitting almost any hypotheses

7. Which of the following statements is not true about the cost function?

The maximum of the concave function is a point where the derivative is 0The maximum of the convex function is a point where the derivative is 0Hill climbing approach is used to find out the minimum or maximum of a cost functionAnalytical Approach is used to find out the minimum or maximum of a cost function

8. What will be the hypotheses outputs for the below NOT function?

f (-20), f (30)f (20), f (-30)f (20), f (-10)f (-20), f (10)

9. Which of the following statement is incorrect about backpropagation?

It is an algorithm commonly used to train the neural networksThe idea behind backpropagation is not to test how wrong the neural network isIt helps to adjust the weights of the neurons so that the accuracy of the output increasesIt is a method of training the neural networks to perform tasks more accurately

10. What is the objective of backpropagation algorithm?

to develop learning algorithm for multilayer feedforward neural networknone of the mentionedto develop learning algorithm for multilayer feedforward neural network, so that network can be trained to capture the mapping implicitlyto develop learning algorithm for single layer feedforward neural network

11. Which of the following Boolean function is represented by the below figure, given that X₁ and X₂ are binary?

ANDORXORNOT

12. Which of the following statements is not true about the given figure?

We can treat a node as an Activation Function, which can output a computation value based on activation functionThe input values can be the input featuresThe terminology of parameters W in NN is called weightsBias unit is mandatory in neural networks

13. The goal of multiclass classification is to construct a function which, given a new data point, will correctly predict the class to which the new point belongs.

FalseTrueDon't knowMaybe

14. Given the below neural network with w₅ = 0.4, Output h₁ = 0.57 and Output h₂ = 0.55. How much does the total net input of O₁ changes with respect to w₅?

0.570.550.450.4

15. A differentiable activation function makes the function computed by a neural network differentiable and the back propagation algorithm is applicable to it.

TrueDon't knowMaybeFalse

16. Mean squared error is a simple and commonly used cost function.

Don't knowFalseTrueMaybe

17. The given classification problem cannot be well solved by using logistic regression.

FalseMaybeTrueDon't know

18. A variational problem is one in which we are looking for a function which can optimize a certain cost function.

Don't knowTrueFalseMaybe

19. What will be the total error of the given neural network given output values O₁ and O₂ are 0.63 and 0.78 respectively?

0.0930.1650.0720.2

20. What will be the final value of f(x, y, z) from the below figure given that x = -5, y = 7, and z = 14?

28-10516-133

21. What are general limitations of back propagation rule?

slow convergencelocal minima problemscalingall of the mentioned

22. When the cost function increases, the learning rate also increases.

FalseDon't knowMaybeTrue

23. Given the below neural network with w₅ = 0.4, Output O₁ = 0.5 and Out h₁ = 0.59. How much does the total error changes with respect to w₅?

0.30.0440.590.25

24. Which of the following statements is not true about the cost function?

It never depends on biasIt is a single value, not a vectorIt is a measure of how good a neural network did with respect to its given training sample and the expected outputIt depend on variables such as weights

25. The cost function of a neural network is always convex.

TrueFalseDon't knowMaybe