第一章 单元测试
1、判断题：
Synapse is the place where neurons connect in function. It is composed of presynaptic membrane, synaptic space and postsynaptic membrane.（ ）
选项：
A:对
B:错
答案: 【对】
2、单选题：
Biological neurons can be divided into sensory neurons, motor neurons and （ ）according to their functions.
选项：
A:multipolar neurons
B:Pseudo unipolar neural networks
C:interneurons
D:bipolar neurons
答案: 【interneurons】
3、判断题：
Neurons and glial cells are the two major parts of the nervous system. （ ）
选项：
A:对
B:错
答案: 【对】
4、判断题：
Neurons are highly polarized cells, which are mainly composed of two parts: the cell body and the synapse. （ ）
选项：
A:对
B:错
答案: 【对】
5、判断题：
The human brain is an important part of the nervous system, which contains more than 86 billion neurons. It is the central information processing organization of human beings. （ ）
选项：
A:错
B:对
答案: 【对】
第二章 单元测试
1、单选题：
In 1989, Mead, the father of VLSI, published his monograph “( )”, in which a genetic neural network model based on evolutionary system theory was proposed.
选项：
A:Perceptrons: An Introduction to Computational Geometry
B:Learning Machines
C:Analog VLSI and Neural Systems
D:Journal Neural Networks
答案: 【Analog VLSI and Neural Systems】
2、判断题：
In 1989, Yann Lecun proposed convolutional neural network and applied it to image processing, which should be the earliest application field of deep learning algorithm. （ ）
选项：
A:错
B:对
答案: 【对】
3、判断题：
In 1954, Eccles, a neurophysiologist at the University of Melbourne, summarized the principle of Dale, a British physiologist, that “each neuron secretes only one kind of transmitter “.（ ）
选项：
A:错
B:对
答案: 【对】
4、判断题：
In 1972, Professor Kohonen of Finland proposed a self-organizing feature map (SOFM) neural network model. （ ）
选项：
A:对
B:错
答案: 【对】
5、判断题：
Prediction and evaluation is an activity of scientific calculation and evaluation of some characteristics and development status of things or events in the future according to the known information of objective objects. （ ）
选项：
A:错
B:对
答案: 【对】
第三章 单元测试
1、判断题：
The function of transfer function in neurons is to get a new mapping output of summer according to the specified function relationship, and then completes the training of artificial neural network. （ ）
选项：
A:对
B:错
答案: 【】
2、判断题：
The determinant changes sign when two rows (or two columns) are exchanged. The value of determinant is zero when two rows (or two columns) are same. （ ）
选项：
A:对
B:错
答案: 【】
3、判断题：
There are two kinds of phenomena in the objective world. The first is the phenomenon that will happen under certain conditions, which is called inevitable phenomenon. The second kind is the phenomenon that may or may not happen under certain conditions, which is called random phenomenon. （ ）
选项：
A:对
B:错
答案: 【】
4、判断题：
Logarithmic S-type transfer function, namely Sigmoid function, is also called S-shaped growth curve in biology. （ ）
选项：
A:对
B:错
答案: 【】
5、判断题：
Rectified linear unit (ReLU), similar to the slope function in mathematics, is the most commonly used transfer function of artificial neural network. （ ）
选项：
A:对
B:错
答案: 【】
第四章 单元测试
1、单选题：
The perceptron learning algorithm is driven by misclassification, so the stochastic gradient descent method is used to optimize the loss function. （ ）
选项：
A:minimum
B:correct
C:maximum
D:misclassification
答案: 【】
2、判断题：
Perceptron is a single-layer neural network, or neuron, which is the smallest unit of neural network. （ ）
选项：
A:错
B:对
答案: 【】
3、判断题：
When the perceptron is learning, each sample will be input into the neuron as a stimulus. The input signal is the feature of each sample, and the expected output is the category of the sample. When the output is different from the category, we can adjust the synaptic weight and bias value until the output of each sample is the same as the category. （ ）
选项：
A:对
B:错
答案: 【】
4、判断题：
If the symmetric hard limit function is selected for the transfer function, the output can be expressed as 
. If the inner product of the row vector and the input vector in the weight matrix is greater than or equal to -b, the output is 1, otherwise the output is -1. （ ）
选项：
A:对
B:错
答案: 【】
5、判断题：
The basic idea of perceptron learning algorithm is to input samples into the network step by step, and adjust the weight matrix of the network according to the difference between the output result and the ideal output, that is to solve the optimization problem of loss function L(w,b). （ ）
选项：
A:错
B:对
答案: 【】
第五章 单元测试
1、单选题：
The output of BP neural network is （ ）of neural network.
选项：
A:the output of the second layer
B:the input of the last layer
C:the input of the second layer
D:the output of the last layer
答案: 【
2、判断题：
BP neural network has become one of the most representative algorithms in the field of artificial intelligence. It has been widely used in signal processing, pattern recognition, machine control (expert system, data compression) and other fields. （ ）
选项：
A:对
B:错
答案: 【】
3、判断题：
In 1974, Paul Werbos of the natural science foundation of the United States first proposed the use of error back propagation algorithm to train artificial neural networks in his doctoral dissertation of Harvard University, and deeply analyzed the possibility of applying it to neural networks, effectively solving the XOR loop problem that single sensor cannot handle. （ ）
选项：
A:错
B:对
答案: 【】
4、判断题：
In the standard BP neural network algorithm and momentum BP algorithm, the learning rate is a constant that remains constant throughout the training process, and the performance of the learning algorithm is very sensitive to the selection of the learning rate. （ ）
选项：
A:错
B:对
答案: 【】
5、判断题：
L-M algorithm is mainly proposed for super large scale neural network, and it is very effective in practical application. （ ）
选项：
A:错
B:对
答案: 【】
第六章 单元测试
1、判断题：
RBF neural network is a novel and effective feedforward neural network, which has the best local approximation and global optimal performance. （ ）
选项：
A:错
B:对
答案: 【】
2、判断题：
At present, RBF neural network has been successfully applied in nonlinear function approximation, time series analysis, data classification, pattern recognition, information processing, image processing, system modeling, control and fault diagnosis. （ ）
选项：
A:对
B:错
答案: 【】
3、判断题：
The basic idea of RBF neural network is to use radial basis function as the “basis” of hidden layer hidden unit to form hidden layer space, and hidden layer transforms input vector. The input data transformation of low dimensional space is mapped into high-dimensional space, so that the problem of linear separability in low-dimensional space can be realized in high-dimensional space. （ ）
选项：
A:错
B:对
答案: 【】
4、判断题：
For the learning algorithm of RBF neural network, the key problem is to determine the center parameters of the output layer node reasonably. （ ）
选项：
A:对
B:错
答案: 【】
5、判断题：
The method of selecting the center of RBF neural network by self-organizing learning is to select the center of RBF neural network by k-means clustering method, which belongs to supervised learning method. （ ）
选项：
A:错
B:对
答案: 【】
第七章 单元测试
1、判断题：
In terms of algorithm, ADALINE neural network adopts W-H learning rule, also known as the least mean square (LMS) algorithm. It is developed from the perceptron algorithm, and its convergence speed and accuracy have been greatly improved. （ ）
选项：
A:对
B:错
答案: 【】
2、判断题：
ADALINE neural network has simple structure and multi-layer structure. It is flexible in practical application and widely used in signal processing, system identification, pattern recognition and intelligent control. （ ）
选项：
A:错
B:对
答案: 【】
3、判断题：
When there are multiple ADALINE in the network, the adaptive linear neural network is also called Madaline which means many Adaline neural networks. （ ）
选项：
A:对
B:错
答案: 【】
4、判断题：
The algorithm used in single-layer ADALINE network is LMS algorithm, which is similar to the algorithm of perceptron, and also belongs to supervised learning algorithm. （ ）
选项：
A:错
B:对
答案: 【】
5、判断题：
In practical application, the inverse of the correlation matrix and the correlation coefficient are not easy to obtain, so the approximate steepest descent method is needed in the algorithm design. The core idea is that the actual mean square error of the network is replaced by the mean square error of the k-th iteration.（ ）
选项：
A:对
B:错
答案: 【】
第八章 单元测试
1、单选题：
Hopfield neural network is a kind of neural network which combines storage system and binary system. It not only provides a model to simulate human memory, but also guarantees the convergence to （ ）.
选项：
A:local maximum
B:minimum
C:local minimum
D:maximum
答案: 【】
2、判断题：
At present, researchers have successfully applied Hopfield neural network to solve the traveling salesman problem (TSP), which is the most representative of optimization combinatorial problems. （ ）
选项：
A:对
B:错
答案: 【】
3、判断题：
In 1982, American scientist John Joseph Hopfield put forward a kind of feedback neural network “Hopfield neural network” in his paper Neural Networks and Physical Systems with Emergent Collective Computational Abilities. （ ）
选项：
A:错
B:对
答案: 【】
4、判断题：
Under the excitation of input x, DHNN enters a dynamic change process, until the state of each neuron is no longer changed, it reaches a stable state. This process is equivalent to the process of network learning and memory, and the final output of the network is the value of each neuron in the stable state. （ ）
选项：
A:错
B:对
答案: 【】
5、判断题：
The order in which neurons adjust their states is not unique. It can be considered that a certain order can be specified or selected randomly. The process of neuron state adjustment includes three situations: from 0 to 1, and 1 to 0 and unchanged. （ ）
选项：
A:对
B:错
答案: 【】
第九章 单元测试
1、判断题：
Compared with GPU, CPU has higher processing speed, and has significant advantages in processing repetitive tasks. （ ）
选项：
A:对
B:错
答案: 【】
2、判断题：
At present, DCNN has become one of the core algorithms in the field of image recognition, but it is unstable when there is a small amount of learning data. （ ）
选项：
A:错
B:对
答案: 【】
3、判断题：
In the field of target detection and classification, the task of the last layer of neural network is to classify. （ ）
选项：
A:错
B:对
答案: 【】
4、判断题：
In AlexNet, there are 650000 neurons with more than 600000 parameters distributed in five convolution layers and three fully connected layers and Softmax layers with 1000 categories. （ ）
选项：
A:错
B:对
答案: 【】
5、判断题：
VGGNet is composed of two parts: the convolution layer and the full connection layer, which can be regarded as the deepened version of AlexNet. （ ）
选项：
A:对
B:错
答案: 【】
第十章 单元测试
1、单选题：
The essence of the optimization process of D and G is to find the（ ）.
选项：
A:local maxima
B:maximum
C:minimum
D:minimax
答案: 【
2、判断题：
In the artificial neural network, the quality of modeling will directly affect the performance of the generative model, but a small amount of prior knowledge is needed for the actual case modeling.（ ）
选项：
A:对
B:错
答案: 【】
3、判断题：
A GAN mainly includes a generator G and a discriminator D. （ ）
选项：
A:错
B:对
答案: 【】
4、判断题：
Because the generative adversarial network does not need to distinguish the lower bound and approximate inference, it avoids the partition function calculation problem caused by the traditional repeated application of Markov chain learning mechanism, and improves the network efficiency. （ ）
选项：
A:对
B:错
答案: 【】
5、判断题：
From the perspective of artificial intelligence, GAN uses neural network to guide neural network, and the idea is very strange. （ ）
选项：
A:错
B:对
答案: 【】
第十一章 单元测试
1、判断题：
The characteristic of Elman neural network is that the output of the hidden layer is delayed and stored by the feedback layer, and the feedback is connected to the input of the hidden layer, which has the function of information storage. （ ）
选项：
A:对
B:错
答案: 【】
2、判断题：
In Elman network, the transfer function of feedback layer is nonlinear function, and the transfer function of output layer islinear function.（ ）
选项：
A:错
B:对
答案: 【】
3、判断题：
The feedback layer is used to memorize the output value of the previous time of the hidden layer unit and return it to the input. Therefore, Elman neural network has dynamic memory function.（ ）
选项：
A:对
B:错
答案: 【】
4、判断题：
The neurons in the hidden layer of Elman network adopt the tangent S-type transfer function, while the output layer adopts the linear transfer function. If there are enough neurons in the feedback layer, the combination of these transfer functions can make Elman neural network approach any function with arbitrary precision in finite time.（ ）
选项：
A:错
B:对
答案: 【】
5、判断题：
Elman neural network is a kind of dynamic recurrent network, which can be divided into full feedback and partial feedback. In the partial recurrent network, the feedforward connection weight can be modified, and the feedback connection is composed of a group of feedback units, and the connection weight cannot be modified. （ ）
选项：
A:错
B:对
答案: 【】
第十二章 单元测试
1、单选题：
The loss function of AdaBoost algorithm is（ ）.
选项：
A:linear function
B:exponential function
C:nonlinear function
D:logarithmic function
答案: 【
2、判断题：
Boosting algorithm is the general name of a class of algorithms. Their common ground is to construct a strong classifier by using a group of weak classifiers. Weak classifier mainly refers to the classifier whose prediction accuracy is not high and far below the ideal classification effect. Strong classifier mainly refers to the classifier with high prediction accuracy. （ ）
选项：
A:错
B:对
答案: 【】
3、判断题：
Among the many improved boosting algorithms, the most successful one is the AdaBoost (adaptive boosting) algorithm proposed by Yoav Freund of University of California San Diego and Robert Schapire of Princeton University in 1996. （ ）
选项：
A:对
B:错
答案: 【】
4、判断题：
The most basic property of AdaBoost is that it reduces the training error continuously in the learning process, that is, the classification error rate on the training data set until each weak classifier is combined into the final ideal classifier. （ ）
选项：
A:对
B:错
答案: 【】
5、判断题：
The main purpose of adding regularization term into the formula of calculating strong classifier is to prevent the over fitting of AdaBoost algorithm, which is usually called step size in algorithm. （ ）
选项：
A:对
B:错
答案: 【】
第十三章 单元测试
1、单选题：
The core layer of SOFM neural network is（ ）.
选项：
A:output layer
B:hidden layer
C:input layer
D:competition layer
答案: 【
2、单选题：
In order to divide the input patterns into several classes, the distance between input pattern vectors should be measured according to the similarity. （ ）are usually used.
选项：
A:Sine method
B:Cosine method
C:Euclidean distance method and cosine method
D:Euclidean distance method
答案: 【
3、判断题：
SOFM neural networks are different from other artificial neural networks in that they adopt competitive learning rather than backward propagation error correction learning method similar to gradient descent, and in a sense, they use neighborhood functions to preserve topological properties of input space. （ ）
选项：
A:对
B:错
答案: 【】
4、判断题：
For SOFM neural network, the competitive transfer function (CTF) response is 0 for the winning neurons, and 1 for other neurons.（ ）
选项：
A:错
B:对
答案: 【】
5、判断题：
When the input pattern to the network does not belong to any pattern in the network training samples, SOFM neural network can only classify it into the closest mode. （ ）
选项：
A:错
B:对
答案: 【】
第十四章 单元测试
1、单选题：
The neural network toolbox contains（ ）module libraries.
选项：
A:four
B:six
C:three
D:five
答案: 【】
2、单选题：
The “netprod” in the network input module can be used for（ ）.
选项：
A:addition or subtraction
B:dot multiplication
C:dot multiplication or dot division
D:dot division
答案: 【
3、判断题：
The “dotrod” in the weight setting module is a normal dot product weight function.（ ）
选项：
A:对
B:错
答案: 【】
4、判断题：
The mathematical model of single neuron is y=f(wx+b).（ ）
选项：
A:错
B:对
答案: 【】
5、判断题：
The neuron model can be divided into three parts: input module, transfer function and output module. （ ）
选项：
A:错
B:对
答案: 【】
第十五章 单元测试
1、判断题：
In large-scale system software design, we need to consider the logical structure and physical structure of software architecture. （ ）
选项：
A:对
B:错
答案: 【】
2、判断题：
The menu property bar has “label” and “tag”. The label is equivalent to the tag value of the menu item, and the tag is the name of the menu display.（ ）
选项：
A:对
B:错
答案: 【】
3、判断题：
It is necessary to determine the structure and parameters of the neural network, including the number of hidden layers, the number of neurons in the hidden layer and the training function.（ ）
选项：
A:错
B:对
答案: 【】
4、判断题：
The description of the property “tooltipstring” is the prompt that appears when the mouse is over the object. （ ）
选项：
A:对
B:错
答案: 【】
5、判断题：
The description of the property “string” is: the text displayed on the object. （ ）
选项：
A:对
B:错
答案: 【】
第十六章 单元测试
1、单选题：
The description of the parameter “validator” of the wx.TextCtrl class is: the（ ）.
选项：
A:position of control
B:size of control
C:style of control
D:validator of control
答案: 【
2、单选题：
The description of the parameter “defaultDir” of class wx.FileDialog is: （ ）.
选项：
A:open the file
B:default file name
C:save the file
D:default path
答案: 【】
3、判断题：
In the design of artificial neural network software based on wxPython, creating GUI means building a framework in which various controls can be added to complete the design of software functions. （ ）
选项：
A:对
B:错
答案: 【】
4、判断题：
When the window event occurs, the main event loop will respond and assign the appropriate event handler to the window event. （ ）
选项：
A:错
B:对
答案: 【】
5、判断题：
From the user’s point of view, the wxPython program is idle for a large part of the time, but when the user or the internal action of the system causes the event, and then the event will drive the wxPython program to produce the corresponding action.（ ）
选项：
A:对
B:错
答案: 【】
