哈尔滨工程大学
- The essence of a perceptron is a linear classification model suitable for classifying some data into two types.Its input is the category of the instance, and the output is the feature vector of the instance.（ ）
- Primary vision is the inverse of the optical imaging problem, which consists of a series of processes that can recover three-dimensional visible object surfaces from two-dimensional light intensities.（ ）
- Deep learning is a learning method based on unsupervised feature learning and feature hierarchy.（ ）
- Neuron is the most basic structural and functional unit of the nervous system, which is divided into two parts: cell body and process. The cell body is composed of nucleus, cell membrane, and cytoplasm; there are two kinds of processes: dendrites and axons.（ ）
- Problem solving technology mainly includes two aspects, namely problem representation and solution method. The state space of a problem is a graph representing all possible states of the problem and their relationships. It contains a set of three descriptions. （ ）
- Cognition is the process in which the human brain processes information received from the outside world and converts it into internal psychological activities, and then controls human behavior. It is the most basic psychological process of human beings.（ ）
- Whether the following statement is correct or not? Robots may be classified as "the third being" in the future, which is different from life/living things (the first being) and non life/living things (the second being). Their external and behavior are similar to living things, but they do not have self-consciousness.（ ）
- The essence of deep learning is to learn more useful features by building a multi hidden layer model and massive training data (which can be unlabeled data), so as to ultimately improve the accuracy of classification or prediction. （ ）
- Swarm robot system has the following three characteristics: robustness, flexibility and scalability. （ ）
- Is the following statement correct? "Smart City" is a new urban development model integrating urban development planning, urban livelihood management and urban industrial development. （ ）
- The cumulative effect of the signals received by a neuron determines the state of that neuron.（ ）
- The depth-first search algorithm stores the visited nodes in a stack.（ ）
- The development of robots has gone through three generations of program-controlled robots (first generation), adaptive robots (second generation), and intelligent robots (modern).（ ）
- The application of computer vision in the field of measurement includes angle measurement and length measurement.（ ）
- An interesting property of Naive Bayes is that it works well for very large datasets.（ ）
- The pattern does not refer to the thing itself but the information obtained from the thing.Therefore, the pattern often appears as distribution information with time and space, so the pattern can be the image of iris, face, fingerprint, or some kind of electrical signal.（ ）
- Machine vision systems can quickly acquire a large amount of information. Although it is difficult to process automatically, it is easy to integrate with design information and process control information.（ ）
- The problem solving system using semantic network to represent knowledge is called semantic network system. There are two main reasoning processes: inheritance and abandonment. （ ）
- Digital image processing system is mainly composed of image digitizing equipment (such as scanner, digital camera, etc.), image processing computer (such as PC) and image output equipment (such as printer, plotter, etc.). （ ）
- In actual classification applications, because the feature vectors often have many dimensions, the classification of label values is preferred to choose the binary classification model。（ ）
- Which of the following statements is correct（ ）.
- What aspects does a computer "understand" natural language?（ ）.
- The following options belong to the characteristics （ ）。
- The following statements about BP network are correct: （ ）。
- Which of the following statements about perception and attention is true（ ）.
- The following belong to the three subsystems of the attention network are（ ）
- The following belong to the essentials of the state-space method:（ ）
- The following statements about the basic characteristics of artificial neural network are correct: （ ）。
- The following are not in the form of knowledge graphs （ ）。
- Which of the following correctly describes supervised learning vs unsupervised learning:（ ）.
- The reason why deep learning adopts the hierarchical network structure of neural network is that（ ）.
- Common methods of transfer learning in the following options are （ ） 。
- The reason why the deep neural network in deep learning generally does not use the BP algorithm is:（ ）.
- The following statements about machine learning are correct: （ ）。
- Among the following options, the English abbreviations of neural network models correspond to the Chinese names: （ ）。
- The process of forming the eigenvector includes:（ ）
- The basic process of an automatic question answering system (sorting questions)（ ）.
- Which of the following statements about the history of artificial intelligence development is correct（ ）.
- Among the following options, the core part of exoskeleton robotics is （ ）。
- Criteria for natural language "understanding": Give a computer a piece of natural language text, if the computer can（ ）.
- Which of the following statements about machine translation is correct（ ）.
- Which of the following statements about the perceptron model is incorrect（ ）.
- The following is not part of the process of computer simulation of human communication: （ ）。
- When designing a BP neural network, The design steps are generally （ ）。(1) Determination of the number of hidden layers and hidden layer neurons(2) Setting of initial weights(3) Pre processing of training data(4) Post processing process
- Which of the following options is not a reason for transfer learning:（ ）
- Which of the following fields of application does machine translation belong to?（ ）
- Which of the following statements about the left and right hemispheres of the brain and their functions is incorrect（ ）.
- Among the following options, those that do not belong to the common methods of environmental modeling are :（ ）。
- The advantages of PSO algorithm applied to path planning are:（ ）.
- The key foundation and technology that the intelligent question answering system mainly relies on do not include the following aspects （ ）。
- The following cells that are not related to animal navigation are（ ）.
- To apply to various polygons, which of the following environment modeling methods should be applied （ ）。
- Among the following options, what does not belong to the basic purpose of computer vision is （ ）。
- Which of the following statements about intelligence is false（ ）.
- Which of the following statements about language and natural language understanding is wrong（ ）.
- The following is incorrect about the development history of machine translation in China（ ）.
- In the __ era of the 20th century, AT&TBell (Bell) Research Institute successfully developed the Audry system. （ ）
- Which of the following statements about supervised learning classification methods is incorrect（ ）
- What is incorrect about machine translation is that （ ）。
- Which of the following coordinate systems expresses the location of a pixel in an image in physical units（ ）.
A:错 B:对
答案:错
A:错 B:对
答案:对
A:对 B:错
答案:对
A:对 B:错
答案:对
A:对 B:错
答案:对
A:对 B:错
答案:对
A:对 B:错
答案:对
A:错 B:对
答案:对
A:错 B:对
答案:对
A:对 B:错
A:对 B:错
A:错 B:对
A:错 B:对
A:错 B:对
A:对 B:错
A:对 B:错
A:错 B:对
A:对 B:错
A:错 B:对
A:错 B:对
A:AI programs think the same way humans do B:For specific tasks, artificial intelligence programs have the ability to learn autonomously C:Artificial intelligence is a computer program that makes reasonable actions based on the perception of the environment and obtains the greatest benefit D:Any computer has artificial intelligence
A:copy B:translate C:question and answer D:Abstract generation
A:Feedback is not timely, there is a delay B:Strong versatility and promotion C:Data serialization, data and data are related D:No need for supervision
A:BP network is a backfeed network B:Basic BP algorithm includes two processes of forward propagation of signal and back propagation of error C:The core three parts of BP neural network learning algorithm are weight adjustment, output layer connection weight adjustment Adjustment of hidden layer connection weight D:Traditional BP network generally uses two-level network
A:Attention is the state of mental activity or consciousness at a certain moment, which is manifested as the direction and concentration of a certain object. B:In the process of perception, there are two basic forms of sensation and perception. C:Perception is produced by the objective world directly acting on human sensory organs. D:The attentional network is divided into three subsystems: pre-attentional attitude, post-attentional attitude and vigilance system.
A:Pre-Attention Mindset B:Post-Attention Mindset C:alert system D:screening system
A:operator B:surgery C:state D:state space method
A:The connection strength between neurons determines the strength of signal transmission B:The cumulative effect of signals received by a neuron determines the state of the neuron C:The signal can be stimulating or inhibiting D:The connection strength between neurons can be changed with training
A:graph database B:RDF C:Semantic Web D:knowledge base
A:Independent vs Non-Independent B:Labeled vs Unlabeled C:Classification vs Clustering D:Classification and qualitative at the same time vs first clustering and then qualitative
A:High-level features or images, often composed of some basic structures (shallow features) B:Hierarchy of vision, where learning of attributes is a compositional mapping that makes categories into attributes C:Structural (or semantic) high-level features are more meaningful for classification D:Structural feature representation can be achieved by using a hierarchical network structure, and different features are represented by different structures
A:Model-based transfer learning B:Sample-based transfer learning C:Feature-based transfer learning D:Relation-based transfer learning
A:Multi-layer network, only the connection between adjacent layer nodes, can not be corrected B:The BP algorithm requires labeled data for training, but most of the data is unlabeled C:Convergence is easy to local minimum. Since it is initialized with random values, it is easy to cause this situation when the initial value is far from the optimal area. D:During feedback adjustment, the gradient becomes more and more sparse, and the error correction signal becomes smaller and smaller from the top layer down
A:The data set for supervised learning has labels. Compared with supervised learning, unsupervised learning has no training process, Instead, we directly take the data for modeling and analysis B:The supervision of supervised learning is reflected in that all data to be processed by machines must be manually defined with corresponding categories, and then the classification algorithm is trained to obtain a classifier that can be used C:According to different learning theories, machine learning models can be divided into supervised learning, semi supervised learning, unsupervised learning, and reinforcement learning, Machine learning is mainly unsupervised learning or semi supervised learning.
A:CNN -- convolutional neural network B:GAN -- generation of confrontation network C:DNN -- deep neural network D:RNN -- cyclic neural network
A:feature evaluation B:feature selection C:feature extraction D:feature formation
A:User query intent analysis and understanding B:Parsing user input C:Query and Retrieval D:Candidate answer generation and ranking
A:In 1986, a parallel computing research group in the United States proposed the backpropagation learning (BP) learning algorithm of the previous feedback neural network. B:In 1957 Rosenblatt defined a neural network structure called a perceptron. C:In 1982, physicist Hopfield proposed a fully connected network, a discrete neural network model. This is a brand new neural network model with a complete theoretical basis. D:In 1949, the psychologist Hebb proposed the Hebb rule in his book "Behavioral Self-Organization".
A:Energy Sources B:Materials of Construction C:Control and Actuation D:Biomechanics
A:Abstract generation B:translate C:paraphrase D:question and answer
A:The basic principle of the statistical machine translation method is to realize the mapping of the source language vocabulary to the target language vocabulary B:In essence, the processing of language by machine translation is to build a bridge between two languages and realize the equivalent conversion from one language to another. C:Statistical-based machine translation methods generally go through four stages when performing machine translation: retrieval, analysis, conversion, and generation. D:The translation method based on intermediate language is to analyze the source language to generate a representation called intermediate language, and then directly generate the target language from the representation of the intermediate language.
A:Perceptrons can solve XOR problems. B:The essence of a perceptron is a linear classification model suitable for classifying some data into two types. C:The basic idea of perceptron learning: the change of connection weight between neural units is proportional to the difference between the expected output and the actual output of the output unit. D:The perceptron model consists of two layers, the input layer and the output layer, without loss of generality.
A:Convert language into corresponding speech B:Convert human voice into language C:Recognize the speech content of the expression language D:Understand the meaning of the language expressed by speech.
A:(1) (2) (3) (4) B:(1) (3) (2) (4) C:(2) (1) (3) (4)
A:The contradiction between big data and strong computing B:The Contradiction Between Universal Models and Personalized Needs C:The contradiction between big data and less labeling D:application-specific needs
A:Simulation of human senses B:expert system C:machine learning D:natural language system
A:When answering questions orally, the left brain is responsible for logical thinking, and the right brain is responsible for language description. B:The right brain is like an artist, with creativity and rich emotions in artistic activities such as music and art. C:The left brain is like a scientist, good at abstract thinking and complex calculations, but lacks rich emotions. D:The two left and right hemispheres of the brain have the same shape, but their functions are quite different. They are generally called the left brain and the right brain respectively.
A:Grid method B:Artificial potential field method C:Visual graph method D:Free space method
A:It is not easy to fall into the local optimal solution and cause deadlock phenomenon B:Small amount of calculation, good real-time performance, simple structure C:Can be combined with other algorithms D:High performance of its own parameters
A:Professional intelligent algorithm B:Strong natural language processing technology
A:head direction cells B:border cells C:speed cell D:cells outside the entorhinal
A:Voronoi diagram method B:grid method C:visual graph method D:free space method
A:Calculate the surface physical properties of the target object according to one or more two-dimensional images. B:Calculate the distance from the observation point to the target object according to one or more two-dimensional images C:To realize the understanding of the three-dimensional scene world, That is to realize some functions of the human visual system D:Calculate the motion parameters of the target object according to one or more two-dimensional images
A:Bacteria have no intelligence. B:Currently, human intelligence is the highest level of natural intelligence. C:From the perspective of life, intelligence is the basic ability of life to adapt to the natural world. D:All life has intelligence.
A:The research goal of natural language understanding is to establish a sufficiently accurate language mathematical model to enable computers to complete natural language related tasks through programming. B:Artificial language refers to machine language, including C++, C language, etc. C:Natural language refers only to spoken or written language used for human communication D:Natural language understanding is a multidisciplinary subject
A:At the end of 1996 Orient Express set off a climax of translation software in China with the slogan of "Smart Sinicization" B:In 1956, my country began to study machine translation. In 1987, the Chinese Academy of Military Sciences successfully developed "Keyi No. 3", which marked that my country's machine translation system has grown from scratch. C:At the end of 1997, Orient Express set off a climax of translation software in China with the slogan of "Smart Sinicization" D:1998 was the most prosperous year for translation software in China. Dozens of translation software such as Tongyi, Yixing, Landau, Instant Messenger, Hanshen, and Internet Collection appeared in China.
A:70 B:50 C:60 D:80
A:The purpose of decision tree learning is to deal with unseen examples of strong decision tree B:The method of support vector machines is suitable for high-dimensional spaces and can be effectively used for small data sets C:Naive Bayes is a method that allows building classifiers in a simple and straightforward manner D:In the KNN nearest neighbor method, a larger K value during classification can reduce the influence of noise and make the boundaries between categories clear.
A:Machine translation of non literary publications has been carrying out product specifications The translation effect of some materials based on professional corpus, such as weather forecast, is good B:The translation of Internet and computer related materials depends largely on machine translation. Machine translation plays an incomparable role in this respect. C:More than 70% of websites in the world are written in English, while only about 2% of websites are written in Chinese. D:Machine translation is the best way to translate science and technology, especially for specific disciplines and majors. Because the sentence structure of this kind of translation is relatively simple, and professional vocabulary has a large number of repetitions, while machine translation is fast and efficient, it can fully play its role in this respect.
A:image coordinate system B:world coordinate system C:image plane coordinate system D:camera coordinate system
温馨提示支付 ￥5.00 元后可查看付费内容,请先翻页预览!
