H13-321_V2.0 HCIP-AI-EI Developer V2.0 Exam Free Dumps

dumps2024-09-23T10:57:41+08:00

H13-321_V2.0 HCIP-AI-EI Developer V2.0 exam is a required one for HCIP-AI-EI Developer certification. There are 60 single-answer questions, multiple-answer questions, true or false, short response items, drag and drop items in the real H13-321_V2.0 exam, and you have 90 minutes to answer all the questions. The required passing score for the Huawei H13-321_V2.0 exam is 600 (out of 1000). The exam cost is $300. The available languages are English and Chinese. We provide the latest HCIP-AI-EI Developer V2.0 H13-321_V2.0 exam dumps, which are valuable for you to pass. Free HCIP-AI-EI Developer V2.0 H13-321_V2.0 exam dumps are available below.

Page 1 of 17

1. What is the abbreviation of the conditional random field algorithm?

SVM

CRF

HMM

GMM

2. Solve equations

4-1-3-2-5

2-1-3-4-5

1-3-4-2-5

2-3-1-4-5

3. Which of the following is not a method of speech feature extraction?

PLP

RNN

MFCC

LPC

4. Which of the following is not one of the three major problems of HMM?

Testing

Evaluation

Training

Decoding

5. Which of the following is not an application of hidden Markov model in speech recognition?

Calculate the output sequence

Use the BW algorithm to find the optimal solution

Phonemes and words combined according to the best sequence

Form words and sentences based on labels

6. What is Gaussian distribution also called?

Random distribution

Discrete distribution

Normal distribution

Abnormal distribution

7. What is the biggest problem with parameter synthesis method?

Poor sound quality

Parameters are difficult to obtain

Uncertain waveform

Speech rate unknown

8. Which of the following is not a speech feature extraction method?

MFCC

LPC

LPCC

SVM

9. Which of the following is correct about the Gaussian distribution curve?

The Gaussian curve is bell-shaped, with high ends and low in the middle.

The smaller the standard deviation, the flatter the curve

The Gaussian curve is bell-shaped, with low ends and high in the middle.

The larger the standard deviation, the taller the curve.

10. Which of the following is not a speech acoustic feature?

Duration

Semantics

Frequency

Amplitude

Page 2 of 17

11. What is the role of Gaussian mixture model in speech recognition?

Represents the probability transfer between factors

Multidimensional numerical distribution used to represent a phoneme

Used to distinguish different phonemes

Evaluating the Advantages of the HMM Algorithm

12. How difficult is speech recognition?

Personal characteristics

Scenarios

Geographical factors

All of the above

13. Which of the following is the most important change in the CD-DNN-HMM model?

Introducing more hidden layers

Introducing Dropout

Introducing the fully connected layer

Introducing contextual information

14. What is the dimension of the commonly used Mel-Frequency Cepstral Coefficients?

15. Which of the following is not a NLP language feature?

Contextual word features

Text Features

Texture features

Single word features

16. Which of the following models can be used for natural language processing?

SVM

HMM

Linear regression

17. Which of the following is not a difficulty in named entity recognition?

Large quantity

Complex rules

Complex nesting

Uniform length

18. Which of the following belongs to the language model?

Adaboost

N-gram

19. Which of the following is not a common method for text vectorization?

DBOW

CBOW

20. What is the unique gate structure of GRU?

Input Gate

Output Gate

Forget Gate

Update Gate

Page 3 of 17

21. Which of the following options does not belong to the three levels of natural language processing technology?

Speech Analysis

Lexical Analysis

Syntactic Analysis

Semantic Analysis

22. Which of the following statements about participles is incorrect?

Rule-based word segmentation is simple and efficient, but dictionary maintenance is difficult.

The purpose of statistical segmentation is to perform probability calculation on the segmentation results and obtain the segmentation method with the highest probability.

Chinese is different from English natural word segmentation. Chinese word segmentation is a basic step in text processing. The performance of word segmentation directly affects the performance of other modules such as part of speech and syntax tree.

In actual engineering applications, word segmentation generally only uses one method.

23. In the NNLM model, what is the activation function used in the network output layer?

Tanh

Softmax

ReLU

Sigmoid

24. Which of the following options does not belong to the gate of the LSTM cell?

Forget Gate

Output Gate

Memory Gate

Input Gate

25. Which of the following is the general format of a URL?

https://Endpoint

https://uri

https://www.XXXX

https://Endpoint/uri

26. Which of the following design specifications does Huawei Cloud API comply with?

Restful

P8EP

Programming

PEP8

27. Which of the following indicators cannot be output by the classification_report in the sklearn.metrics model evaluation indicator module?

f1-Score

recall

accuracy

precision

28. When using the image tagging API provided by Huawei Cloud, which field in the response parameter is the image tag content?

labels

content

result

tags

29. How can the entire digital image after sampling and quantization be represented mathematically?

Channel

Matrix

Coordinates

Pixels

30. In which year did Huawei launch the Ascend series IP and processors?

2018

2016

2019

2017

Page 4 of 17

31. Which of the following options is the algorithm used to estimate the parameters of a single Gaussian model?

EM algorithm

Maximum Likelihood Estimation

Viterbi algorithm

Forward-backward algorithm

32. When using ModelArts automatic learning to build a predictive analysis project, if the label column is enumerated data, which of the following is the label column data type and the model type to be trained?

Continuous Values and Regression Models

Discrete Values and Regression Models

Continuous Numerical and Categorical Models

Discrete Values and Categorical Models

33. Which of the following options is a method for training static word vectors?

Fasttext

ELMO

BERT

GPT

34. Which of the following options does not belong to the category of phonetics?

Acoustic Phonetics

Linguistic Phonetics

Auditory Phonetics

Articulation and Phonetics

Semantic Phonetics

35. When using ModelArts automatic learning to build a predictive analysis project, if the label column is numerical continuous data, which of the following is the label column data type and the model type to be trained?

Discrete Values and Regression Models

Continuous Numerical and Categorical Models

Discrete Values and Categorical Models

Continuous Values and Regression Models

36. Which of the following options is not a sequence annotation method?

CRF

MEMM

N-Gram

HNM

37. Which of the following is not a field supported by the intent understanding API provided by Huawei Cloud EI natural language processing service?

Music

Alarm clock

Jokes

Entertainment

38. For a pixel P with coordinates (x, y), P has four horizontal and vertical neighboring pixels, called a 4-neighborhood. Which of the following can correctly represent the coordinates of the 4-neighborhood?

(x-1, y), (x-1, y), (x, y-1), (x, y+1)

(x-1, y), (x+1, y), (x, y+1), (x, y+1)

(x-1, y), (x+1, y), (x, y-1), (x, y-1)

(x-1, y), (x+1, y), (x, y-1), (x, y+1)

39. In the image preprocessing filtering experiment, which of the following functions calls OpenCV's median blur API?

cv2.GaussianBlur(im, (5, 5), 0)

cv2.blur(im, (3, 3))

cv2.medianBlur(im, 5)

cv2.filter2D(im, -1, sharpen_1)

40. Which step does not belong to speech recognition preprocessing?

Deduplication

Noise Reduction

Silent Cutoff

Standardization

Page 5 of 17

41. Which of the following options does not belong to the three elements of the Markov model?

State Collection

Observation Probability

Transition Probability

Initial Probability

42. If the sampling frequency is f and the number of frames is a, which of the following is the length of the sound signal?

a/2f

2a/f

a/f

a*f

43. Which of the following is the abbreviation of speech recognition?

TTS

ASR

STT

AST

44. Which of the following is the abbreviation of speech synthesis?

TTS

ASR

STT

AST

45. In the Hidden Markov Model, which of the following options can be calculated using the Gaussian Mixture Model (GMM)?

Language Probability

Transition Probability

Observation Probability

Initial Probability

46. Regarding CBOW and Skip-Gram algorithms, which of the following statements is incorrect?

Both CBOW and Skip-Gram can be used to train word embeddings, but Skip-Gram is faster than CBO

CBOW calculates the probability of a word appearing based on the n words preceding it or the n consecutive words before and after it.

Skip-Gram calculates the probabilities of several words before and after a certain word.

Both the CBOW model and the skip-gram model are based on the Huffman tree.

47. Which of the following options is not a commonly used algorithm for keyword extraction?

TextRank

SSA

LDA

TF-IDF

48. Which of the following options is not a method in the development of natural language processing?

Statistical-based methods

Rule-based approach

Deep Learning-Based Methods

Recursion-based methods

49. The bag-of - words model is the earliest text vectorization method that uses words as the basic processing unit.

Which of the following options is not a defect of the bag-of-words model?

Unable to preserve order information

Based on the distribution hypothesis

The curse of dimensionality

There is a semantic gap

50. In which of the following fields does the convolutional neural network perform well?

Computer Vision

Speech Recognition

Machine Translation

Knowledge Graph

Page 6 of 17

51. Which of the following schools of thought does the expert system belong to?

Symbolism

Connectionism (Neural Networks)

Behaviorism (robot dogs, robot animals)

None of the above

52. A car company's products, factory A accounts for 60%, factory B accounts for 40%, the defective rate of A is 1%, the defective rate of B is 3%. Now a car is selected and it is defective. What is the probability that it was produced by factory A?

1/2

1/3

2/3

1/6

53. If a model has a small deviation and a large variance on the test set, which of the following does it belong to?

Underfitting

Overfitting

Exact fit

It may be overfitting or underfitting

54. Which of the following options is the type of classification task label?

Discrete

Continuous

Automorphic

Strain type

55. How does the learning rate change when using the AdaGrad algorithm?

Reduce

Increase

Increase first and then decrease

Decrease first and then increase

56. The model is trained based on the labeled data. The output of the model prediction is a continuous value. An example of this type of task is to predict the claim amount of the insured (used to set the insurance premium). What type of task does this machine learning belong to?

Regression

Clustering

Generate

Classification

57. In the process of model training and evaluation, overfitting and underfitting often occur. Which of the following descriptions of overfitting and underfitting is correct?

Overfitting is reflected in the evaluation indicators, that is, the model performs well on the training set, but performs poorly on the test set and new data.

Underfitting refers to the situation where the model performs well during training but performs poorly during prediction.

Overfitting refers to a situation where the model does not fit the training data sufficiently.

Underfitting refers to a situation where the model performs well both during training and prediction.

58. In the field of deep learning, which of the following activation functions have a value range between 0 and positive infinity?

SeLU

sigmoid

tanh

ReLU

59. Which of the following scientists created the Lisp programming language?

Alan Turing

Kilfond Shannon

Marvin Minsky

John McCarthy

60. In machine learning, how many support vectors are there for the SVM algorithm?

Several

Page 7 of 17

61. Which of the following options makes GPU suitable for accelerating the operation of neural networks?

Control and Scheduling Capabilities

Parallel computing capabilities

Serial computing capability

The ability to call data from memory

62. What determines the number of neurons in the input layer of a neural network?

Data Dimensions

Number of data items

Number of categories

Number of parameters

63. Which of the following options is not an integrated learning algorithm in machine learning?

Bagging

Boosting

Stacking

Marking

64. Which of the following statements about artificial neural networks is correct?

Artificial neural network is a network composed of interconnected artificial neurons

Artificial neural network is an abstraction and simplification of the human brain from the perspective of microstructure and function. It is an important way to simulate human intelligence.

Artificial neural networks reflect several basic characteristics of human brain functions, such as parallel information processing, learning, association, pattern classification, memory, etc.

All of the above are correct

65. The back propagation algorithm is an important method for updating parameters of neural networks. The back propagation algorithm is implemented with the help of which of the following calculation methods?

Computational Graph

Higher-order differentials

Cost Function

Chain Rule

66. Which of the following concepts is used to calculate the derivative of a composite function?

Chain Structure in Calculus

Hard hyperbolic tangent function

softplus function

Vibration basis function

67. In the vehicle detection scenario of traffic law enforcement, it is better to miss the penalty for the illegal vehicle than to report it falsely, so as to avoid the wrong fine. Which indicator of the system needs to be very high?

Recall

Precision

Accuracy

Confidence

68. Which function is not included in the back propagation algorithm?

Incentive Communication

Weight Update

Normalization

Get the gradient

69. The structure of the perceptron simulates which of the following biological structures?

Brain

Neurons

Eyes

Ears

70. What is the purpose of using gradient descent?

Finding label values

Changing the model structure

Optimization parameters

Prediction results (forward calculation in neural network)

Page 8 of 17

71. What is the main difference between supervised and unsupervised algorithms?

Data volume

Validation Set

Tags

Test set

72. What problem does the ReLU activation function mainly solve compared to the Sigmoid function?

There is no derivative at coordinate 0

Segmentation is not smooth

Monotonically increasing

Gradient disappearance

73. What layer is the Softmax function mainly used for?

Input layer

Hidden Layer

Output layer

74. What is the main function of the convolution kernel?

Dimensionality reduction

Extracting local features

Color Enhancement

Change image size

75. What is the main function of the fully connected layer?

Classification

Extracting features (convolutional layer)

Dimensionality reduction (pooling layer)

Change image size (image scaling, downsampling, upsampling)

76. What is the advantage of stochastic gradient descent compared to batch gradient descent?

Easier to converge (stochastic gradient descent convergence is more unstable)

Faster iteration speed

Higher accuracy

Stronger generalization ability

77. Which statistical analysis icon is used to analyze the bottleneck of user conversion?

Daily Active Users

Collapse Trend

Channel Analysis

Path Funnel Model

78. Which of the following is not a component of the DNN structure?

Hidden Layer

Weight

Activation Function

Dropout

79. Which of the following problems cannot be solved by a single-layer perceptron?

Simple binary classification problem

OR Problem

AND Problem

XOR problem

80. Which of the following options is not an algorithm used in the Hidden Markov Chain Evaluation Problem?

Baum-Welch algorithm

Exhaustive method

Forward Algorithm

Backward Algorithm

Page 9 of 17

81. Which of the following is correct about the difference between CNN applications in text and image tasks?

CNN is suitable for image-related tasks, but performs poorly on text tasks

CNN performs poorly on named entity recognition tasks because the features it extracts are similar to the N-gram model, which is the local relationship between words.

The input of color images is single-channel, and the input of text is multi-channel

CNN performs poorly on text classification tasks because the features it extracts are similar to the N-gram model, which is the local relationship between words.

82. In an 8-bit RGB color image, what color does (255, 255, 255) represent?

Blue

Red

Black

White

83. When taking a photo of a document with a mobile phone, the document in the image will be distorted due to the shooting angle. You can find the coordinates of () corresponding key points, perform () transformation on the image, and correct the document image.

4, affine

4, Perspective

3, Perspective

3, affine

84. Which of the following is not a common feature?

HOG

Haar

LBP

Ground Truth

85. The target in the image is 15 pixels to the right of the center. Which of the following translation coordinate transformations can be used to move the target to the center of the image?

(0, 15)

(0, -15)

(15, 0)

(-15, 0)

86. How to label image classification tasks?

Use a circle to frame the target

Use an ellipse to frame the target

Use a rectangle to frame the target

Enter a category label for the image

87. When taking a selfie, what transformation does the image displayed by the front camera of the mobile phone need to undergo to become a real scene image?

Translation

Rotation

Vertical Mirror

Horizontal mirror

88. The brightness of an image refers to how bright the image is. How many variations can a 4-bit image distinguish?

128

256

89. When the coordinates change, which of the following methods can we use to get the integer value of the pixel?

Sampling

Mean

Quantification

Interpolation

90. Calculate the LBP feature of a pixel in a 3*3 window. If the value of the pixel is smaller than the surrounding pixels, the calculated feature value is?

255

127

256

Page 10 of 17

91. Image recognition tasks can be divided into three levels, from low to high according to the abstractness of the processing content?

Image analysis, image processing, image understanding

Image analysis, image understanding, image processing

Image processing, image analysis, image understanding

Image understanding, image analysis, image processing

92. Use binarization to segment character regions. Due to noise, all character connected regions are stuck together. What method can be used to separate the characters?

Opening

Expansion

Closing operation

Gamma Correction

93. Use a 3*3 convolution kernel to convolve a three-channel color image. The number of parameters that need to be learned is?

94. Take a photo indoors on a cloudy day with poor lighting. Calculate the brightness histogram of the photo. What shape might it be?

The left side is relatively flat, and the right side has a peak. (The image is brighter)

The right side is relatively flat, and the left side has a peak. (The image is relatively low in brightness)

There is a peak on each side

The waveform is uniform and has no peaks

95. When taking photos against the light, the face is relatively dark. Use gamma correction to enhance the image to achieve a clear effect even in backlight. What value of γ can be selected?

0.5

96. What is the output of graphical analysis?

Image size (no need to analyze, can be directly retrieved)

Image Content

Image pixel value (no need to analyze, can be directly retrieved)

Image grayscale value (no need to analyze, can be directly retrieved)

97. What problems of mean filtering can Gaussian filtering overcome?

Denoising (the main function of Gaussian filtering is noise reduction)

Fuzzification

Slow calculation (the calculation principle is the same, so the calculation speed depends on the data size)

Large amount of calculation (same as C)

98. What is the main basis for image classification?

Color

Shape

Features

Image size

99. When paying by face recognition, which of the following steps is correct?

Image preprocessing - Image recognition - Face detection

Face recognition - image preprocessing - image detection

Image preprocessing - object detection - face recognition

Image preprocessing - face recognition - face segmentation

100. Gamma change is () processing, histogram equalization is () processing, mean filtering is () processing. ()

Point processing, local processing, and overall processing

Local processing, point processing, and overall processing

Point processing, overall processing, local processing

Overall processing, point processing, local processing

Page 11 of 17

101. A photo has many small noise points during storage. The best denoising effect is to perform () operation on the scanned copy.

Median Filter

Gaussian filtering

Mean Filter

Laplace filtering

102. The facial features detection API detects facial features of the input image and returns the coordinates of key points representing the location of the facial features. How many landmark points are currently supported?

103. What is the spatial adjacency relationship between the two pixels with coordinates (21, 13) and (22, 12)?

No adjacency exists

Diagonal adjacency

8-adjacency

4-adjacency

104. Many mobile phones provide eye protection screen display mode, which can reduce blue light and relieve fatigue. This function can be achieved by adjusting the image ()

Brightness

Saturation

Contrast

Hue

105. The OTSU algorithm traverses possible thresholds and selects the maximum segmentation threshold between the foreground area and the background area () as the optimal threshold.

Absolute difference

Area difference

Cumulative difference of pixel values

Between-class variance

106. What will change in the image if the H component of the HSV color space is modified?

Hue (warmth or coldness of the image)

Brightness (V)

Saturation (S)

Contrast

107. Use two 3x3 convolution kernels to convolve a three-channel color image. How many channels does the resulting feature map have?

108. We often talk about the 720P video format. What does 720 represent?

Video frame rate

Image resolution height

Width of image resolution

Grayscale of the video

109. Which of the following templates can achieve image sharpening?

0 -1 0 -1 4 -1 0 -1 0

0 1 0 1 3 1 0 1 0

0 -1 0 -1 6 -1

0 1 0 1 1 1 0 1 0

110. Through image algorithms, the foreground and background of a person in a photo taken with an ordinary mobile phone are segmented, and then () operations are performed on () parts to achieve a blur effect similar to that of a SLR camera.

Background, median filter

Background, mean filter

Foreground, mean filter

Foreground, median filter

Page 12 of 17

111. In the figure below, the 4-connected path from pixel p to t is ()

P u

Q

RVT

S

pqrvt

puqrst

puqrvt

pqrst

112. In the figure below, the chessboard distance between pixels p and s is () (& symbol represents a space)

P

&q

& r&t

&&&s

113. Which of the following steps is not included in the image preprocessing process?

Image Enhancement

Image Recognition

Grayscale transformation

Geometric Correction

114. For a 32-level grayscale image, how many levels can be distinguished from the darkest to the brightest at each sampling point?

128

256

115. What color does the RGB value (255, 0, 0) represent?

Pure blue

Pure black

Pure green

Pure red

116. What color does the RGB value (124, 252, 0) represent?

Sky blue

Pure black

Pure blue

Lawn Green

117. For a pixel P with coordinates (x, y), P has four diagonally adjacent pixels, called diagonal neighbors. Which of the following can correctly represent the coordinates of the diagonal neighbors?

(x-1, y-1), (x-1, y+1), (x+1, y-1), (x+1, y+1)

(x-1, y-1), (x-1, y), (x+1, y-1), (x+1, y+1)

(x-1, y), (x-1, y+1), (x+1, y), (x+1, y+1)

(x-1, y), (x-1, y+1), (x+1, y-1), (x+1, y)

118. The V component of the HSV color space represents the brightness (value) of the color. What is the value range?

0 to 1

0 to 100

1 to 100

0 to 99

119. Which of the following neural networks contains feedback connections?

Convolutional Neural Network

Recurrent Neural Network

Convolutional Generation Network

Deep Belief Network

120. Which activation function is used by the forget gate in LSTM?

Sigmoid

Tanh

ReLu

Softmax

Page 13 of 17

121. N-gram is defined as N keywords grouped together. How many triple phrases can be generated from a given sentence?

[Analytics Vidhya is a great source to learn data science].

122. Which of the following descriptions of the shortcomings of the co -occurrence matrix + SVD is wrong?

SVD only learns shallow semantics (correct)

The co-occurrence matrix does not take into account the order of words (correct)

SVD calculation is large

The co-occurrence matrix is high-dimensional and sparse (correct)

123. Which of the following options does not belong to the understanding and analysis of natural language?

Structural Analysis

Syntactic Analysis

Lexical Analysis

Voice Analysis

124. When trying to use SelectKBest for feature selection, which of the following options uses the card method to select features?

chi2

f_regression

mutual_info_classif

mutual_info_regression

125. Huawei Cloud EI Enterprise Intelligence provides a variety of efficient services. Which of the following is the correct specific process for users to use the services?

Apply for service - Call API - Obtain request authentication - View service usage information

Get request authentication - Apply for service - Call API - View service usage information

Get request authentication - Apply for service - View service usage information - Call API

Apply for service-obtain request authentication-call API-view service usage information

126. Which of the following options is the architecture adopted by Huawei’s Atlas series chips?

Von Neumann

Gaussian

Ascend

Leonardo da Vinci

127. Which of the following options is the application classification of Al chip business?

Inference and acceleration

Acceleration and Training

Training and Inference

Compilation and training

128. Which of the following options is Huawei’s artificial intelligence computing platform?

ModelArts

Huawei Cloud EI

ECS

Atlas

129. Which of the following is the highlights of Huawei’s deep learning framework MindSpore?

By implementing the Al algorithm as code, the development environment becomes more friendly, significantly reducing model development time and lowering the model development threshold.

Through MindSpore 's own technological innovation and the collaborative optimization of MindSpore and Ascend processors, efficient running state is achieved and computing performance is greatly improved.

Natively adapt to every scenario including terminal, edge and cloud, and can collaborate on demand, with flexible deployment.

All of the above

130. Which of the following types of AI chips does the Ascend AI processor belong to?

NPU

ASIC

FPGA

GPU

Page 14 of 17

131. Which of the following is not the basic computing resource in the AI Core of the Yiteng AI chip?

Matrix calculation unit

Vector calculation unit

Scalar calculation unit

Tensor calculation unit

132. Which of the following is the power consumption of the Yiteng 310 chip?

133. What kind of applications can the predictive analysis function in ModelArts automatic learning be used for?

Predict weather temperature changes

Identify flower species

Segmenting objects in images

Face recognition

134. In which field of a ModelArts training job can user -defined hyperparameters be set?

Resource Pool

Operating parameters

Number of computing nodes

Algorithm source

135. What kind of applications can the predictive analysis function in ModelArts automatic learning be used for?

Identify flower species

Segmenting objects in images

Face Recognition

Predicting house prices

136. Which part of the deep learning development process does the ModelArts data management function correspond to?

Data Preparation

Code debugging

Reasoning

Training

137. What is the essence of ModelArts preset models?

Reinforcement Learning

Transfer Learning

Reinforcement Learning

GAN

138. What is the role of ModelArts development environment?

Training

Prediction

Code Prediction

Data preparation

139. Among the functions provided by ModelArts, which of the following services does online service belong to?

Model Management

Deployment

Model training

Data processing

140. Which of the following annotation types is not available when creating image datasets in ModelArts data management?

Image Segmentation

Image Classification

Object Detection

Image Generation

Page 15 of 17

141. When using ModelArts automatic learning to build image classification or object detection projects, which of the following image formats does the dataset not support?

png

jpg

webp

bmp

142. How to label object detection tasks in data management?

Use a circle to frame the target

Use a rectangle to frame the target

Use an ellipse to frame the target

Enter a category label for the image

143. What is the usage process of ModelArts automatic learning?

Data annotation -> deployment -> model training

Data annotation -> model training -> deployment

Deployment -> Model training -> Data labeling

Model training -> data annotation -> deployment

144. What types of image annotations does ModelArts support?

Image Classification

Object Detection

Image Segmentation

All of the above are true

145. What text annotation tasks does ModelArts support?

Text Classification

Name recognition

Text triples

All of the above are true

146. Which of the following is the API used in mox.file to copy a file?

mox.file .copy

mox.file .rename

mox.file .copyparallel

mox.file .remove

147. Which of the following is the operation performed by the code mox.file.exists (' obs :// bucket_name /test01')?

Create a folder

Read a file on OBS

List an OBS path

Check if the folder exists

148. Which of the following is the API used in mox.file to move an OBS file or folder?

mox.file .rename

mox, file.copyparallel

mox.file .copy

mox.file .remove

149. When creating a training job, which of the following computing resources is not supported by ModeIArts?

GPU

TPU

CPU

Ascend

150. In Huawei Cloud Content Detection Service, which of the following parameters is not a common request message header?

POST

HOST

Authorization

X- Sdk -Date

Page 16 of 17

151. Which of the following options is the API corresponding to the speech synthesis interface?

POST v1.0/voice/ ttl

GET v1.0/voice/ ttl

GET v1.0/voice/ tts

POST v1.0/voice/ tts

152. When using gensim to implement word2vec, which parameter represents the dimension of the output word vector?

min_count

size

window

153. What is Tensorboard?

Deep learning visualization tool officially provided by MXNet

ModelArts’ self-developed visualization tools

Visualization toolkit provided by Python

Deep learning visualization tool provided by TensorFlow

154. What is the function of MoXing's automatic hyperparameter search function?

Provide guidance on manual parameter adjustment

Automatically search for the optimal hyperparameters in the parameter space

Automatically search for the optimal learning rate

Automatically search for the optimal Batch_size

155. Which neural network is used in the face recognition case?

VGG

LSTM

Inception

Resnet

156. TensorFlow 2 does not perform graph operations by default. Which method should be used to use graph operations?

session

tensor_to_dense

tensorbord

@tf.function

157. What operation does "isinstance" represent in TensorFlow2.0?

Determine whether it is numpy

View device information

Determine whether it is a Tensor

Merge tensors

158. Which of the following options is not a feature of ModelArts?

Support Tensorflow and Pytorch

Support model management

Support automatic tuning

Multiple built-in models

159. Which optimizer is used in the following code?

Adadelta = tf.keras.optimizers.Adadelta ()

Model.compile （ loss=' binary_crossentropy ',

Optimizer = ' adam ',

Metrics = ['accuracy'])

Adam

Adadelta

SGD

Adaboost

160. Huawei Cloud EI provides a Restful API. Which of the following options is a method for requesting the server to add new resources or perform special operations?

GET

POST

PUT

Attach

Page 17 of 17

161. Which of the following OpenCV functions can achieve image size scaling?

cv2.resize(img, (200, 300))

cv2.imread （ 1ena.jpg ）

print (type(im ))

cv2.warpAffine(img , M, (w, h)

162. Huawei Cloud EI provides a RESTful API. Which of the following options is a method for requesting the server to return a specified resource?

GET

PUT

Attach

POST

163. Which of the following parameters can be set by the nperseg parameter in the spectrogram function?

Sampling frequency

Frame Shift

Window Function

Frame length

164. When calling Huawei cloud computing voice service, you can call the underlying interface encapsulated by SDK, and you can access the Restful service by simulating the front-end browser. The authentication methods used by the two are?

AK/SK, Token

Token, AK/SK

Token, Token

AK/SK, AK/SK

165. Which of the following is an inference server provided by Shengteng Computing to its customers and partners?

Atlas 800 model

Atlas 300 model

Atlas 800 model

166. REST observes the entire network from the perspective of resources. The resources distributed everywhere are determined by which of the following?

URI (Uniform Resource Identifier)

Restful API

URL (Uniform Resource Locator)

PEP8

167. Which of the following is not an advantage of the ReLU series of activation functions over the Sigmoid and Tanh activation functions?

The unilateral inhibition of ReLU provides the sparse expression ability of the network.

The non-saturation of ReLU can effectively solve the problem of gradient disappearance and provide a relatively wide activation boundary.

ReLU can cause neuron death during its training process.

From a computational perspective, both Sigmoid and Tanh activation functions require exponential calculations and are highly complex, while ReLU only requires a threshold to obtain an activation value.

168. Which of the following descriptions about the bag-of-words model and TF-IDF are correct? (Multiple choice)

The bag-of-words model cannot distinguish the importance of common words and proper nouns to the text

Both the bag-of-words model and TF-IDF ignore the position information of words

Features after bag-of-words vectorization or TF-IDF vectorization usually have the problem of high-dimensional sparsity.

TF-IDF cannot distinguish the importance of common words and proper nouns to the text

H13-321_V2.0 HCIP-AI-EI Developer V2.0 Exam Free Dumps

H13-321_V2.0 HCIP-AI-EI Developer V2.0 Exam Free Dumps

Share this post

Leave a Reply Cancel reply

Related Posts