Free HCIA-AI V3.5 Exam H13-311_V3.5 Exam Practice Test

"Today's speech processing technology can achieve a recognition accuracy of over 90% in any case." Which of the following is true about this statement?

• This statement is incorrect. The accuracy of speech recognition is high, but not extremely high.
• This statement is incorrect. In many situations, noise and background sound have a huge impact on speech recognition accuracy.
• This statement is correct. Speech processing can achieve a high level of accuracy.
• This statement is correct. Speech processing has a long history and the technology is very mature.

In MindSpore, mindspore.nn.Conv2d() is used to create a convolutional layer. Which of the following values can be passed to this API's "pad_mode" parameter?

• pad
• same
• valid
• nopadding

When using the following code to construct a neural network, MindSpore can inherit the Cell class and rewrite the init and construct methods.

• TRUE
• FALSE

Huawei Cloud ModelArts provides ModelBox for device-edge-cloud joint development. Which of the following are its optimization policies?

• Hardware affinity
• Operator optimization
• Automatic segmentation of operators
• Model replication

Which of the following are callback options provided by MindSpore?

• SummaryCollector
• TrainStep
• ModelCheckpoint
• LossMonitor

Which of the following are feedforward neural networks?

• Fully-connected neural networks
• Recurrent neural networks
• Boltzmann machines
• Convolutional neural networks

When feature engineering is complete, which of the following is not a step in the decision tree building process?

• Decision tree generation
• Pruning
• Feature selection
• Data cleansing

The global gradient descent, stochastic gradient descent, and batch gradient descent algorithms are gradient descent algorithms. Which of the following is true about these algorithms?

• The batch gradient algorithm can solve the problem of local minimum value.
• The global gradient algorithm can find the minimum value of the loss function.
• The stochastic gradient algorithm can find the minimum value of the loss function.
• The convergence process of the global gradient algorithm is time-consuming.

Which of the following are use cases of generative adversarial networks?

• Photo repair
• Generating face images
• Generating a 3D model from a 2D image
• Generating images from text

The derivative of the Rectified Linear Unit (ReLU) activation function in the positive interval is always:

• 0.5
• 1
• Variable