HCIA AI practice exam 1

Question 1

1.) What are the element of AI?

Select one or more:

a. Scenarios
b. Algorithms
c. Data
d. Computing power

Answer 1

Computing Power
Algorithms
Scenarios
Data

Question 2

2.) Speech processing is a general term for various processing technologies used to research the voicing process, statistical features of speech signals, speech recognition

True
False

Answer 2

True

Question 3

3. Which of the following belong to Howard Gardner(human intelligence)?

a.Language
b. Space
c.Logic
d. Music

Answer 3

Space, Language, Music, Logic

Question 4

4. What are the application scenarios of computing vision?

a. Smart album
b. Action analysis
c. Voice navigation
d. Image search

Answer 4

Action analysis, image search, smart album

Question 5

5. The three phases of AI include computingg intelligence, (), and cognitive intelligence.

Select one:

Behavior Intelligence
Strong AI
Perceptual Intelligence
Narrow AI

Answer 5

Perceptual Intelligence

Question 6

6. Neural networks are not a type of machine learning

True
False

Answer 6

False

Question 7

7. Which of the following is not integrated learning policy in machine learning algorithms?

Select one:
Stacking
Bagging
Marking
Boosting

Answer 7

Marking

Question 8

8. In machine learning, which of the following is not a part of the process from obtaining data to officially putting the data into a model?

Select one:
Data standardization
Data cleansing
Data visualization
Data dimension deduction

Answer 8

Data visualization

Question 9

9. The test error keep decreasing as model complexity increases

True
False

Answer 9

True

Question 10

10. If an enterprise has a large amount of sales data without labels, which of the following models are suitable for the enterprise to identify VIP customers?

Select one or more:
SVM
Logistic Regression
K-Means
Hierarchical clustering

Answer 10

Hierarchical clustering, k-means

Question 11

11. Which of the following are common clustering algorithms

Select one or more:

Spectral Clustering
K-means
Hierarchical clustering
Density-based clustering

Answer 11

K-means, Spectral clustering, Density-based clustering, Hierarchical clustering

Question 12

12. Which of the following problems can be resolved by using machine learning algorithms?

Select one or more:
Reinforcement problem
Clustering problem
Classification problem
Regression problem

Answer 12

Clustering problem, Classification problem, regression problem

Question 13

13. Which of the following is an evaluation indicator of regression algorithms?

Select one:
Mean squared error
Accuracy
Confusion matrix
Recall

Answer 13

Mean squared error

Question 14

14. The naive Bayes algorithms does not require sample feature to e independent and identically distributed.

Select one:
True
False

Answer 14

False

Question 15

15. What is the correct overall process of machine learning?

Answer 15

Data collection
Data cleansing
Feature extraction and selection
Model Training
Model evaluation and test
Model deployment and integration

Question 16

The F1 value for evaluation classification models is the harmonic mean of indicators including ().

Select one or more:

Precision
Accuracy
Recall
Validity

Answer 16

Recall, Precision

Question 17

Which of the following points constitute support vectors in the SVM without considering regular expressions?

Select one:

Points of a certain class
Points on the separating hyperplane
Points nearest to the separating hyperplane
Points farthest from the separating hyperplane

Answer 17

Points nearest to the separating hyperplane

Question 18

18. About the following code: SVR(kernel-‘rbf’, C=1e3, gamma= 0.01), it means that we use linear kernel functions.

Select one:
True
False

Answer 18

False

Question 19

19. A loss function reflects an error between a target output and an actual output of a neural network. Which of the following is a common loss function in deep learning?

Select one:
Logarithmic loss function
Mean square loss function
Hinge loss function
Exponential loss function

Answer 19

Mean square loss function

Question 20

20. If each later of a neural network uses the Sigmoid activation function and the number of the layer is large, which of the following problems will occur?

select one:
Gradient explosion problem
Overfitting problem
Vanishing gradient problem
Underfitting problem

Answer 20

Vanishing gradient problem

Question 21

21. Perceptron is the simplest neural network among all deep learning neural networks. Which of the following statements about the perceptron’s structure is correct?

Select one:

The perceptron uses the ReLU activation function
The perceptron has no hidden layer
The perceptron uses the Sigmoid activation function
The perceptron has two hidden layers

Answer 21

The perceptron has no hidden layer

Question 22

22. Which of the following methods can be used to resolve data imbalance problems in deep learning tasks?

Select one or more:

Batch deletion
Synthetic sampling
Random oversampling
Random undersampling

Answer 22

Synthetic sampling, Random oversampling, Random undersampling

Question 23

23. Convolutional neural networks are widely used in the field of image processing. Which of the following is not a part of a convolutional neural network?

Select one:
Pooling layer
Fully-connected layer
convolutional layer
Bidirectional hidden layer

Answer 23

Bidirectional hidden layer

Question 24

24. Currently, generative adversarial network are widely used. Which of the following scenarios can use such networks?

Information retrieval
Image generation
Semantic segmentation
Data augmentation

Answer 24

All of the above

Question 25

25. An activation function is used to convert linear mapping into nonlinear mapping. True False

Answer 25

True

Question 26

26. Overfitting means that a model performs well on a training set but poorly on a test set. Which of the following methods can be used to avoid overfitting? Select one or more: Early stopping of training Dropout method L1 and L2 regularization Dataset expansion

Answer 26

all of the above

Question 27

27. How should samples in real datasets be used in Generative Adversarial Networks (GAN)? Select one: Used as the input value of the generative model Used as the output value of the generative model Used as the output value of the discriminative model Used as the input value of the discriminative model

Answer 27

Used as the input value of the discriminative model

Question 28

28. Regularization is an important and effective technology for reducing generalization errors in traditional machine learning. Which of the following are regularization technologies? Select one or more: L1 regularization L2 regularization Momentum optimizer Dropout

Answer 28

Dropout, L1 regularization, L2 regularization

Question 29

29. Which of the following optimizers can automatically adjust the learning rate? Select one: SGD algorithm Momentum optimizer Mini-batch gradient descent algorithm AdaGrad optimizer

Answer 29

AdaGrad optimizer

Question 30

30. As an important part of a neural network, activation functions are classified into various types. Which of the following functions are activation functions Select one or more: ReLU function Sigmoid function SoftPlus function Tanh function

Answer 30

all of the above

Question 31

31. Which of the following statements about neural networks is incorrect? Select one: a. As hidden layers of a neural network increase, the model classification capability gradually weakens. b. The limitation of a single-layer perception is that it cannot resolve XOR problems. c. A feedforward neural network can be represented using a directed acyclic graph d. The neurons at the same layer of the feedforward neural network are not interconnected

Answer 31

a.

Question 32

32. Which of the following statements about a convolutional neural networks is incorrect? Select one: a. During image processing, image convolutional is performed through window scanning. b. a convolutional neural network can include convolutional layers, pooling layers, and fully-connected layers c. a convolutional kernel cannot be used to extract global features of images. d. Common pooling layers include the max-pooling layer and average-pooling layer.

Answer 32

c. A convolutional kernel cannot be used to extract global features of images

Question 33

33. All convolutional kernels of the same covolutional layer on a convolutional neural network share a weight True False

Answer 33

False

Question 34

34. Which of the following operators is not supported by TensorFLow 2.0? Select one: // pow (^) (wa ni parenthesis dma kita gd) @

Answer 34

(^)

Question 35

35. tf.keras.datasets can be used to view Keras built-in datasets. True False

Answer 35

True

Question 36

36. Which of the following steps belong to TensorFLow development process? Select one or more: Data preparation Network construction Model restoration and invoking Model training and verification

Answer 36

all of the above

Question 37

37. Which of the following are built-in loss function of tf.keras.losses? Select one or more BInary cross entropy loss Likelihood Mean square error Mean absolute percentage error loss

Answer 37

Binary cross entropy loss, Mean square error, Mean absolute percentage error loss

Question 38

38. Which of the following methods can be used for tensor combination in TensorFlow 2.0? Select one: Join Split concat unstack

Answer 38

concat

Question 39

39. Which of the following methods is used for network compilation when the Keras interface of TensorFlow 2.0 is used to build a neural networks? Select one: join fit write compile

Answer 39

compile

Question 40

40. Which of the following does not support dimension conversion in TensorFlow 2.0? Select one: reshape squeeze gather transpose

Answer 40

gather

Question 41

41. The graph and session mechanism is deleted from TensorFlow 2.0 True False

Answer 41

True

Question 42

42. which of the following are AI training and inference frameworks? Select one or more: MindSpore PyTorch TensorFlow Pandas

Answer 42

PyTorch, TensorFlow, MindSpore

Question 43

43. The typical application development process of the application development service does not include data loading. True False

Answer 43

False

Question 44

44. Which of the following is not a common operation of MinsSpore? Select one: signal nn array math

Answer 44

signal

Question 45

45. MindSpore can use only Huawei-developed Ascend processors for training and inference. Select one: True False

Answer 45

False

Question 46

46. Which of the following technology innovations is not used by MindSpore to bridge the huge gap between industry research and all-scenario AI applications and facilitate inclusive AI? Select one: New programming paradigm New programming language New way of cooperation New execution paradigm

Answer 46

New programming language

Question 47

47. Bonus

Answer 47

bonus

Question 48

48. The CPU improves AI performance by adding instruction and cores. True False

Answer 48

True

Question 49

49. Which of the following computing resources are included in the Da Vinci architecture? Select one or more: Scalar unit Cube unit Vector unit Tensor unit

Answer 49

Cube, Vector, Scalar

Question 50

50. The GPU is good at processing easy-to-parallel programs with intensive computing Select one: True False

Answer 50

True

Question 51

51. There are multiple models of Atlas 800 AI servers. Which of the following are based on the Kunpeng processor platform? Select one or more: Atlas 800 (model 3000) Atlas 800 (model 3010) Atlas 800 (model 9000)

Answer 51

Atlas 800(model 9000), Atlas 800 (model 3000)

Question 52

52. What are the advantages of the HIAI mobile computing platform? Select one or more: Source code enabling quick start Various tool chains Comprehensive documents Different types of APIs

Answer 52

all of the above

Question 53

53. Which of the following functions is provided by Huawei hiai foundation module? Select one: Enable services to automatically target and acquire users Integrate apps quickly convert and port existing models Push services based on user requirements

Answer 53

Quickly convert and port existing models

Question 54

54. HiAI Engine opens app capabilities and integrates multiple AI capabilities into apps, making apps smarter and more powerful. What can apps benefit from Huawei HiAI? Select one or more: Lower cost Security Ready-to-use Stability

Answer 54

all of the above

Question 55

55. Which of the following scenarios can Enterprise Intelligence( EI) be applied to? Select one or more: Smart City Smart government Smart manufacturing Smart finance

Answer 55

all of the above

Question 56

56. Which of the following services belong to the Huawei EI service family? Select one or more: Natural language processing Basic EI services EI big data services Conversational bot service (CBS)

Answer 56

all of the above

Question 57

57. Which of the following are correct statements about the product advantages of the ModelArts AI development platform? Select one or more: One-stop platform Flexibility Easy of use High performance

Answer 57

All of the above

Question 58

58. Which of the following is not the application of the image recognition service? Select one: Scenario analysis Smart Album Speech synthesis Object detection

Answer 58

Speech synthesis

Question 59

59. Which of the following services can be used together with ModelArts to easily deploy models on devices? Select one Hilens ECS OBS OCR

Answer 59

HiLens

Question 60

60. Which one doesnt belong to huawei EI Essential Platform? Select one: ModelArts Huawei HiLens Huawei HiAI Graph Engine Service (GES)

Answer 60

Huawei HiAI