WorksheetsHCIA-AI part 1 (1-4)
Total questions: 42
Worksheet time: 25mins
Which of the following are sub-fields of AI?
Machine learning
Computer Vision
Speech recognition
Natural language processing
Voice recognition refers to recognizing audio data as text data
TRUE
FALSE
The three stages of artificial intelligence include computational intelligence, () and cognitive intelligence.
Weak artificial intelligence
Perceptual Intelligence
Behavioral Intelligence
Strong artificial intelligence
Which of the following options does not belong to the scope of Huawei's full-stack solution?
Application enable
Edge computing
Open source framework
Chip enable
What are the elements of artificial intelligence?
Algorithm
Scene
Computing power
Data
In the neural network based on connectionism, each node can express a specific meaning.
TRUE
FALSE
Where is the foothold of symbolism?
The foothold is in neuron networks and deep learning
The foothold is in reasoning, symbolic reasoning and machine reasoning
The foothold is perception and action
The foothold is behavior control, adaptive and evolutionary computing.
Which of the following are the application areas of AI?
Wisdom Education
Smart City
Smart Home
Smart medica
Which of the following schools does neural network research belong to?
Symbolism
Connectionism
Behaviorism
None of the above
What are the main technical applications of artificial intelligence?
Natural language processing
Control system
Computer Vision
Speech recognition
Grid search is a method of parameter adjustment
True
False
The current stage of artificial intelligence is still in the weak artificial intelligence stage?
True
False
Artificial intelligence is the research and development of theories, methods and application systems for simulating, extending and expanding human intelligence Of a new technological science?
True
False
Which of the following statements about supervised learning is correct?
Decision tree is a supervised learning
Supervised learning cannot use cross-validation for training
Supervised learning is a rule-based algorithm
Supervised learning can be trained without labels
When dealing with actual problems, when should machine learning be used?
The data distribution itself changes over time and requires continuous re-adaptation of the program, such as predicting the trend of merchandise sales.
The complexity of the rules is low and the problem is small.
The rules of the task will change over time, such as the production line. Defect detection on
The rules are very complicated or cannot be described, such as face recognition and speech recognition
Regularization is an important and effective technique for reducing generalization errors in traditional machine learning. The following techniques are regular The technology is:
L1 regularization
L2 regularization
Dropout
Momentum optimizer
Which of the following is correct about the difference between machine learning algorithms and traditional rule-based methods?
Traditional rule-based methods, in which the rules can be explicitly clarified manually
Traditional rule-based methods use explicit programming to solve problems
The mapping relationship of the model in machine learning is automatically learned
The mapping relationship of the model in the machine learning institute must be implicit
When we describe housing, we often use the attributes of residential area, house type, decoration type, etc. If we use plain With Bayes as a model, we assume that there is no relationship between attributes
True
False
Loss function and model function are the same thing
True
False
In polynomial regression, there is a square term in the formula of the model, so it is not linear.
True
False
What are the common types of dirty data?
Malformed value
Duplicate value
logically wrong value
Missing values
What is the most important difference between batch gradient descent, small batch gradient descent, and stochastic gradient descent?
Gradient size
Gradient direction
Learning rate
Number of samples used
Which of the following evaluation indicators belong to the regression algorithm?
Recall rate
Confusion matrix
Mean square error
Accuracy
In the random forest, what strategy does the final ensemble model use to determine the model result?
Cumulative system
Find the average
Voting system
Multiplying system
The test error will keep getting smaller as the complexity of the model increases
True
False
What are the common clustering algorithms?
Density clustering
Hierarchical clustering
Spectral clustering
Kmeans
In supervised learning, which of the following models is used to describe "the one who is close to the red and the one who is red to the black"?
K-Means
SVM
KNN
Neural network
The naive Bayes algorithm does not require independent and identical distribution among sample features
True
False
Feedforward neural network is a simple neural network, each neuron is arranged hierarchically. It is currently one of the most widely used and fastest growing artificial neural networks. Which of the following statements about feedforward neural networks are correct:
Neurons with computing power are connected to the upper and lower layers B.
Its input nodes have computing capabilities
Connecting neurons in the same layer
Information between layers is only transmitted in one direction
In the neural network, which of the following methods are used to update the parameters when training the network to minimize The loss function?
Forward propagation algorithm
Pooling calculation
Convolution calculation
Backpropagation algorithm
The optimizer is an important part of training neural networks. The purpose of using the optimizer does not include which of the following:
Speed up algorithm convergence
Reduce the difficulty of setting manual parameters
Avoid overfitting --> normalizer
Avoid local extremes
Can the perceptron be displayed in space?
Line
Plane
Hyperplane
Point
Weight sharing is common in neural networks. Which of the following neural networks will share weights?
Perceptron
Convolutional Neural Network
Fully connected neural network
Recurrent neural network
In deep learning tasks, when encountering data imbalance problems, which of the following methods can we use to solve the problem?
Batch delete
Random oversampling
Synthetic sampling
Random undersampling
Regarding backpropagation, is the following statement wrong?
Backpropagation can only be used in feedforward neural networks
Backpropagation can be combined with gradient descent algorithm to update network weights
Backpropagation will go through the activation function
Back propagation refers to the back propagation of errors through the network
What are the commonly used loss functions in deep learning?
L1 loss function
Mean square error loss function
Cross entropy error loss function
In the process of deep learning model training, what are the common optimizers?
Adam
Adagrad
SGD
Momentum
In the classic convolutional neural network model, what hidden layer does the Softmax function follow?
Convolutional layer
Pooling layer
Fully connected layer
All of the above
What is wrong about the following statement about cyclic neural network?
Cycle The neural network can be expanded according to the time axis
LSTM cannot solve the problem of gradient disappearance
LSTM is also a recurrent neural network
Recurrent neural network can be abbreviated as RNN
In the deep learning network, the backpropagation algorithm is used to find the optimal parameters. What rules are used in the backpropagation algorithm to obtain the derivation layer by layer?
Chain rule
The rule of accumulation
Law of Equivalence
Rule of Normalization
L1 and L2 regularization are methods commonly used in traditional machine learning to reduce generalization errors. The following is about the two The correct method is:
L1 regularization can be used for feature selection
Both L1 and L2 regularization can be used for feature selection
L2 regularization can do feature selection
Neither L1 nor L2 regularization can be used for feature selection
Traditional machine learning and deep learning are the core technologies of artificial intelligence, and there is a slight difference in the engineering process. The following steps What you don't need to do in deep learning is
Model evaluation
Feature Engineering
Data cleaning
Model construction
