Which of the following statements are A. True about commonly used optimizers?

The idea behind the Adagrad optimizer Is to set different learning rates for different parameters

One drawback of the Adagrad optimizer is that it ends the optim ization process too early

The momentum optimizer updates parameters with the same learning rate, while the momentum coefficient is adjusted with each iteration.

The RMSProp optimizer introduces an attenuation coefficient to enable the gradients to attenuate by a certain proportion in each round

Which of the following statements are True about the dropout regularization?

It is more effective than penalty parameter in deep learning

It is simple to calculate and easy to implement

Its effect is poor with insufficient training data

Generally, it is used only in training

Which of the following statements are A. True about the advantages of the Rectified Linear Unit (ReLU) activation function?

The vanishing gradient problem can be effectively alleviated

The output is bounded, meaning the training is not easy to diverge.

Which of the following statements are True about model parameters and hyperparameters?

Models contain both parameters and hyperparameters

Hyperparameters are manually set

Hyperparameters can be used to control training

Hyperparameters are automatically learned by models

Which of the following statements about data preprocessing are True?

Data cleansing Is a process of filling in missing values, as well as detecting and eliminating noise data and exceptions.

The purpose of data dimension reduction ts to simplify data attributes and avoid the curse of dimensionality.

The purpose of data standardization is to reduce noise data and improve model accuracy by standardizing data.

Machine learning outputs results through models. Therefore, model training is more important than data preprocessing.

Which of the following statements about datasets are True?

A dataset is a collection of data used in machine learning tasks. Each piece of data is called a sample.

Events or attributes that reflect the performance or nature of a sample in a particular aspect are called features

A training set is a dataset used in the training process, where each sample is referred to as a training sample.

Learning (or training) is the process of building a model from data.

Which of the following statements about some of Al use case are True?

Smart home uses technologies such as the Internet of Things (loT) and speech recognition

Smart healthcare uses Al technologies such as com puter vision and data mining.

Smart city is a comprehensive domain, covering all three basic technology fields of Al: speech, NLP, and com puter vision

Smart education features openness, sharing, interaction, collaboration, and ubiquity. It aims to modernize education and upgrade traditional teaching methods with information technologies

Which of the following statements about the three major schools of Al are True?

Symbolicism puts more focus on logical inference. Connectionism is relatively weak in terms of being logical and interpretability. Instead, it focuses on results.

Connectionism stems from the study of bionics, especially human brain models

Actionism puts more focus on practice. It believes in constantly adapting actions by learning from the environment

Symbolicism puts more focus on practice. It believes in constantly adapting actions by learning from the environment

Which of the following statements about the L1 buffer is True?

The L1 buffer decreases data accesses over the bus and avoids bus congestion

Data in the L1 buffer needs to be read outside the Al Core over the bus interface each time.

The L1 buffer can permanently retrain data that needs to be reused

L1 buffer is used to store the initial values in the neural network.

Which of the following statements about the Da Vinci architecture is incorrect?

Acompute unit contains four types of basic compute resources

Control units are responsible for the running of Al Cores

The storage system consists of on-chip storage units of Al Core and corresponding data paths

Transfers data to the L1 buffer through the bus interface unit.

On-device execution refers to the execution of the entire graph. It makes full use of the computing power of the Ascend Al Processor, greatly reducing the interaction overhead and improving the accelerator usage. Which of the following is false about on-device execution?

The chip-oriented deep graph optimization is used to reduce synchronization waiting time and maximize the parallelism degree of 'data-computing communication". The training performance is equivalent to that of the graph scheduling mode on the host.

Challenges to model execution with powerful chip com puting power: Memory wall problems, high interaction overhead, and difficult data supply. Some operations are performed on the host, while others are performed on the device. The interaction overhead is much greater than the execution overhead. As a result, the accelerator usage is low

Challenges to distributed gradient aggregation with powerful chip com puting power: When a single iteration ¢ ResNet-50 takes 20 ms, the central control overhead and communication overhead for frequent synchronization are generated. The traditional method requires three times of synchronization to complete AllReduce. The data-driven method automatically performs AllReduce without control overhead.

MindSpore uses gradient-driven adaptive graph optimization to implement decentralized and autonomous All Reduce. The gradient aggregation step is consistent, and computing and communication are fully streamlined.

Which of the following is NOT a complexity feature of Al computing?

Parallel processing of structured and unstructured data

Parallel com munication and computing

Which of the following statements about the running process of the MindArmour subsystem is false?

Fuzzing execution: Generate trusted test data randomly based on the model coverage and configuration policies

Configuration policies: Define test policies based on threat vectors and trustworthiness certification requirements and select appropriate test data generation methods

Evaluation report: Generate an evaluation report based on built-in or user-defined trustworthiness metrics

Trustworthiness enhancement: Use preset methods to enhance the trustworthiness of Al models

Which of the following are false about convolutional neural networks?

Convolution kernels cannot extract global features of images

A convolutional neural network may contain convolutional, pooling, and fully connected layers

Common pooling includes max pooling and average pooling.

When an image is processed, convolution is implemented by using a scanning window

Mock Exam 1

Quiz

•

Computers

•

University

•

Practice Problem

•

Medium

Irma Dewi

Used 3+ times

FREE Resource

62 questions

Show all answers

1.

FILL IN THE BLANK QUESTION

45 sec • 1 pt

In Huawei Cloud NLP service, the ( ) technology can divide text into separate words as a sequence. For example, in English text, spaces are natural delimiters between words. (Fill in the blank.)

2.

FILL IN THE BLANK QUESTION

45 sec • 1 pt

Among the MindSpore's Python APIs, mindspore.( ) defines loss functions, optimizers, and computing units for constructing networks.

3.

FILL IN THE BLANK QUESTION

45 sec • 1 pt

CNN is an abbreviation for ( ) (use a lowercase)

4.

FILL IN THE BLANK QUESTION

45 sec • 1 pt

At a convolution layer of a convolutional neural network, assume there are 128 3 x 3 convolution kernels and the size of the input feature map is 28 x 28 x 64. The depth of a convolution kernel is ( ). (Enter only digits.)

5.

FILL IN THE BLANK QUESTION

45 sec • 1 pt

The feature interpretability in deep learning is ( ) than in traditional machine learning. (Enter "stronger" or "weaker".)

6.

FILL IN THE BLANK QUESTION

45 sec • 1 pt

In the CART algorithm, the classification tree measures the purity mainly through the ( ) coefficient.

7.

MULTIPLE CHOICE QUESTION

45 sec • 1 pt

Compared with CPUs and GPUs, NPUs use synapse weights to integrate storage and compute, improving operating efficiency.

True

False

Access all questions and much more by creating a free account

Create resources

Host any resource

Get auto-graded reports

Continue with Google

Continue with Email

Continue with Classlink

Continue with Clever

or continue with

Microsoft

Apple

Others

Already have an account?

Similar Resources on Wayground

63 questions

test mask

Quiz

•

University

60 questions

E-Commerce Online Midterm Exam

Quiz

•

University

60 questions

Jaringan Komputer

Quiz

•

University

60 questions

HTML & CSS

Quiz

•

University

59 questions

ICT_Test

Quiz

•

University

60 questions

TIN 7 Bài 1,2,3

Quiz

•

7th Grade - University

60 questions

AI-900 Azure AI Fundamentals

Quiz

•

University

60 questions

Bab4-Jaringan Komputer dan Internet -Gr5

Quiz

•

4th Grade - University

Popular Resources on Wayground

15 questions

Fractions on a Number Line

Quiz

•

3rd Grade

20 questions

Equivalent Fractions

Quiz

•

3rd Grade

25 questions

Multiplication Facts

Quiz

•

5th Grade

54 questions

Analyzing Line Graphs & Tables

Quiz

•

4th Grade

$fractions$

22 questions

fractions

Quiz

•

3rd Grade

20 questions

Main Idea and Details

Quiz

•

5th Grade

20 questions

Context Clues

Quiz

•

6th Grade

15 questions

Equivalent Fractions

Quiz

•

4th Grade

Discover more resources for Computers

20 questions

CompTIA Network+ - Ports and Protocols

Quiz

•

University

Mock Exam 1

62 questions

In Huawei Cloud NLP service, the ( ) technology can divide text into separate words as a sequence. For example, in English text, spaces are natural delimiters between words. (Fill in the blank.)

Among the MindSpore's Python APIs, mindspore.( ) defines loss functions, optimizers, and computing units for constructing networks.

CNN is an abbreviation for ( ) (use a lowercase)

At a convolution layer of a convolutional neural network, assume there are 128 3 x 3 convolution kernels and the size of the input feature map is 28 x 28 x 64. The depth of a convolution kernel is ( ). (Enter only digits.)

The feature interpretability in deep learning is ( ) than in traditional machine learning. (Enter "stronger" or "weaker".)

In the CART algorithm, the classification tree measures the purity mainly through the ( ) coefficient.

Compared with CPUs and GPUs, NPUs use synapse weights to integrate storage and compute, improving operating efficiency.

The HMS Core supports HarmonyOS and Android, but does not support Windows.

Huawei Cloud El provides intelligent twins, Al development platforms, and general Al capabilities.

Products of Atlas series cover training and inference scenarios, including servers, edge computing devices, and acceleration modules.

Access all questions and much more by creating a free account

Similar Resources on Wayground

Popular Resources on Wayground

Discover more resources for Computers