‘A. The value of the hyperparameter is not ned by the algorithm itself. B. Most machine learning algorithms have hyperparameters. OC. typerparameters are parameters that set values before the algorithm begi D.Hyperparameters cannot be modified ‘QUESTION NO: 62 Numerical calculation refers to the method and process of effectively using the digital computer to solve the approximate problem of mathematical problems, and the discipline consisting of related theories. Which ofthe following processes are involved in solving actual problems with a computer? (Multiple Choice) CA. Programming BB. Know the actual problem . Mathematical model © D. Computer calculation results E. Numerical ealeulation method (QUESTION NO: 53, |Which of the following can improve the computational efficiency of the neural network model? (Multiple choice) AFPGA B. Large-scale distributed cluster c.GPu 0 p.TPU ‘QUESTION NO: 64 Which ofthe following is not MindSpore Features of the core architecture? ‘A. Automatic differentiation B. Automatic parallel Automatic coding © D, Automatic tuning QUESTION No: 65 Huawel's chip support HUAWE! HIAI Which module of AHIAI Framework © B.HIAI Service © €.HIAI Foundation D.HIAI Engine‘QUESTION NO: 56 What are the numeric types of Python? (Multiple Choice) A float (floating point type) Bint (integer type) OC. complex (complex number type) 2 D Leng (long integer type) ‘QUESTION No: 67 When using TensorFlow2. 0 of keras When building « neural network with an interface, the network needs to be ‘compiled, Which ofthe following methods should be used to work? > A.join > B. write c.fit D.compile QUESTION NO: 58 Regarding the convolutional neural network pooling ayer, the following description is correct?” ‘A. The pooled feature image becomes smaller B. Pooling operation Is realized by scanning window . Commonly used pooling methods include maximum pooling and average pooling D. The pooling layer can play a role in dimensionality reduction ‘QUESTION NO: 69, Global gradient descent algorithm, stochastic gradient descent algorithm and batch gradient descent algorithm are all gradient descent algorithms. The following is wrong about its advantages and disadvantages. © A. The convergence process ofthe global gradient algorithm is time-consuming © 8. Batch gradient algorithm can solve the local minimum problem © €. Stochastic gradient algorithm can find the minimum value of the loss function D. The global gradient algorithm can find the minimum value of the loss function ‘QUESTION NO: 60 Huawei Al The full scenarios include public cloud, private cloud, various edge computing, loT industry terminal land consumer terminals and other end, edge, and cloud deployment environments, © ATRUE BLFALSE‘QUESTION No: 61 Deep learning neural network training requires a lot of matrix calculations. Generally, we need to use hardware to ‘enable the computer to have parallel computing capabilities. The following hardware devices can provide parallel ‘computing capabilities. [A Motherboard B.GPU ) CRAM D.cPu QUESTION NO: 62 It the deep learning neural network has the problem of gradient disappearance or gradient explosion, our common solution is. ‘A. Random undersampling B. use Relu Activation function C. Gradient shear © D. Regulerization QUESTION NO: 63, ‘When training the network, we often encounter many problems. For the problem of gradient disappearance, we can alleviate the problem by choosing which ofthe following functions to use? A. Softsign function © B.tanh function © €, Sigmoid function D. Relu function QUESTION NO: 64 In a neural network based on connectionism, each node can express a specific meaning, A TRUE BL FALSE ‘QUESTION NO: 65 Which ofthe following data types does Tensortlow not support? A. doubless B.ftoati2 ©. boo! > D. inteQUESTION NO: 68 Which ofthe following items are included in the results returned when the face called? ch service is successfully © A. Searched face position B, Searched face number © ©. Searched taces ia © D. Searched face similarity QUESTION No: 67 Which ofthe following services long to Huawei El Service family? 0 A.E1 Basic services OB Natural language processing ( C.E1Big data service OB. ialogue robot ‘QUESTION NO: 68 Regularization Is an important and effective technique for reducing generalization errors in traditional machine learning. The following techniques are regular. The technology is: CA Dropout (0 B.L2 Regularization 0 C.L4 Regularization OD. Momentum optimizer ‘QUESTION NO: 69 When dealing with actual problems, when should machine learning be used in the following situations? (CA. The complexity ofthe rules Is low, and the problem is small ‘Task rules will change over time, such as defect detection on the produetion line CC. The rules are very complicated or cannot be described, such as face recognition and voice recognition CD. The data distribution itself changes overtime, requiring continuous re-adaptation of the program, such a predicting the trend of merchandise sales QUESTION NO: 70, ‘Which of the following is the computing model of TenserFlow? © A. Conversation B.Caleulation chart ©. Tensor > D. Variable‘QUESTION NO: 71 ‘TensorFlow2.0 not support tensor in GPU with CPU Transfer between each other. A.TRUE ) B FALSE (QUESTION NO: 72 Artificial intelligence is a new technical sclence that studies and develops theories, methods and application systems for simulating, extending and extending human intelligence. Its one of the core research areas of machine learning. A False B.True ‘QUESTION NO: 73, Generating a confrontation network is ike a game system. The generator generates fake samples, discriminator judges whether itis true or false. Our ideal result is? (© A. The discriminator cannot distinguish the true and false samples generated by the generator © B. The generator produces diferent samples . The generator produces roughly the same samples D. The discriminator efficiently distinguishes the true and false samples generated by the generator ‘QUESTION NO: 74 When the voice recognition service is successtully called, which field is the recognition result stored in? © A.content > B. result © c.text D. data ‘QUESTION NO: 75 Which ofthe following conditions is not a condition that n-fold bernoull tris needs to meet? (© A. There are only two possible outcomes for each tial, |e. event A occurs and event A does not occur. B. Each test was repeated under the came conditions. (© €. Each test itself is obeying normal dietribution (© D. The results of each trial are independent of each other.QUESTION NO: 78 Which ofthe following Is not a module in the Tensorflow library? A theontib B. tbeost ©. tt layers D.tinn QUESTION NO: 7 Which of the following statements about universal form recognition services are correct? ‘A. colums Represents the column information occupied by the text block, the number is from 0 Start, list form © B. words Representative text block recognition result C €. The incoming image data needs to go through basesé coding D. rows Represents the line information occupied by the text block, the number s from 0 Start, list form QUESTION No: 78, Which ofthe following is not an artificial intelligence school? > A. Behaviorism B. Symbolism ©. Statsttealtsm > D. Connectionism ‘QUESTION NO: 79, ‘The for loop statement in the Python language can iterate through the items in any sequence. A False B.Tue ‘QUESTION No: 80 ‘In machine learning, what input the model needs to train itset and predict the unknown? A. Neural Networks B historical data Training algorithm D. Manual procedure ‘QUESTION No: 81 Tensorflow supports mult/-TPU cluster computing A False BTue‘QUESTION NO: 82 AIChip Is also called Al Accelerators are specially designed to handle a large number of computing tasks in artifical inteligence applications functional module. A TRUE 8. FALSE ‘QUESTION No: 83, ‘What are the common types of dirty data? ‘A. Missing value B. Malformed value Duplicate value D. Logically wrong value QUESTION NO: 84 What are the commonly used activation functions? (Multiple Choice) O A tanh Brelu ©. sigmoid D. danish ‘QUESTION No: 85 Which of the following description about the relationship between the Poisson distribution and the binomial distribution is correct? [A Poisson distribution can replace binomial distribution © B. Poisson dtstnbutJon ts an approxtmalion of the bmomtal distnbubon when n ts very large and p 1 small ( > ©, The mathematical model of the Poisson distribution and the binomial distribution are both Lagrangian models. © 0, Poisson distribution has no relationship with binomial distribution ‘QUESTION NO: 88 What is wrong description ofthe Python module? A. The Python module statements. ‘Python file that ends with py and contains Python object definitions and Python © B. Python modules can define functions classes and variables but the module does not contain executable code cas Janing relevant code to a module can make your code better and easier to understand. . The Python module allows you to logically organize your Python code snippets.‘QUESTION NO: 87 ‘TensorF1ow Which ofthe following companies frst developed? A.Nvid B. Facebook © €.Geogte OD. Oracte ‘QUESTION NO: 88 center 32°32 Image with size 66 The step size of the convolution kernel ist Convolution calculation, output image Sizes: 28°28 25°23 28°23 > D.29°28 QUESTION No: 89 What are the steps that are not part ofthe operation of a Python file object? A.read © B.write E.close QUESTION No: 90 ‘The Python language dees not allow embedding another loop body in a loop body > A.False Tue ‘QUESTION NO: 91 Which of the following statements about Python are correct? (Multiple choice) |A. Python 13 purely free software and the source code follows the GPL (GNU General Public License) protocol 8. Python syntax is simple and clean One of the festures isto force blank characters to be used ‘indentation [Right Answers) statement C. Invented m 1989 by the Dutch Guido van Rossum, the first public release was issued in 1991. CB. Python 1s often nicknamed glue language which can easily connect various modules made tn other languages‘QUESTION NO: 92 ‘TensorFlow 2.0 stand by GPU The acceleration is mainly manifested inthe parallel operation of the addition, subtraction, multiplication and division of the matrix accelerate. ‘ACTRUE BL FALSE ‘QUESTION NO: 93, ‘Traditional machine learning and deep learning are the core technologies of artifical intelligence. There is a slight lifference in the engineering process. The following steps. What you don’t need to do in deep leaming is © A. Data cleaning Feature engineering © 6. Mosel butising . Mode! evaluation ‘QUESTION NO: 94 ‘The number or matrix is stored in the tensor A False B.True ‘QUESTION NO: 95 agging In integrated learning, the relationship between each base learner is? O A. Additive relationship B. Independent (© ©. The latter model must be built on top of the previous model D. relationship QUESTION NO: 96 Which of the following is not part of the desp learning development framework? A Keras © B.eNTK ©. SAFA > D. MKNet‘QUESTION NO: 97 Where should the labeled data be placed in the confrontation ger > AAs input value for generative model © B.As the output value ofthe discriminant mode! €.As the input value ofthe discriminant model .As the output value of the generated mode! (QUESTION NO: 98 ‘According to the development process of the robot, tis usually divided into three generations, respectively (Multiple Choice) 1 A. Robot with sensation oe aching Reproduction Robot } €. Intelligent robot ) D. Robots that will think ‘QUESTION NO: 99, ‘Which of the following statements about passport recognition services are correct? ‘A: nationality Nationality of representative holder B. country_code Country code representing the country where the passport was issued C. confidence Confidence information of related flelds. The greater the confidence, the higher the reliability of the corresponding field identified thi time. Ina statistical sense, the greater the confidence, the higher the accuracy D. passport_number Representative passport number QUESTION NO: 100, Cut off 2019 At the end of the year,TensorFlow The released versions are? At Boa cs D2