CN117610734A - Deep learning-based user behavior prediction method, system and electronic equipment - Google Patents
Deep learning-based user behavior prediction method, system and electronic equipment Download PDFInfo
- Publication number
- CN117610734A CN117610734A CN202311648970.8A CN202311648970A CN117610734A CN 117610734 A CN117610734 A CN 117610734A CN 202311648970 A CN202311648970 A CN 202311648970A CN 117610734 A CN117610734 A CN 117610734A
- Authority
- CN
- China
- Prior art keywords
- historical
- data
- training
- feature
- traffic network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
- 230000006399 behavior Effects 0.000 title claims abstract description 47
- 238000013135 deep learning Methods 0.000 title claims abstract description 43
- 238000000034 method Methods 0.000 title claims abstract description 43
- 239000011159 matrix material Substances 0.000 claims description 143
- 238000012549 training Methods 0.000 claims description 105
- 239000013598 vector Substances 0.000 claims description 45
- 238000010586 diagram Methods 0.000 claims description 20
- 230000007704 transition Effects 0.000 claims description 14
- 238000000605 extraction Methods 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 13
- 238000012545 processing Methods 0.000 claims description 10
- 238000011176 pooling Methods 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 2
- 238000009826 distribution Methods 0.000 description 8
- 238000013528 artificial neural network Methods 0.000 description 7
- 230000008859 change Effects 0.000 description 6
- 238000007726 management method Methods 0.000 description 6
- 238000003062 neural network model Methods 0.000 description 6
- 238000013473 artificial intelligence Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 230000001965 increasing effect Effects 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000013439 planning Methods 0.000 description 3
- 238000001556 precipitation Methods 0.000 description 3
- 208000009119 Giant Axonal Neuropathy Diseases 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 201000003382 giant axonal neuropathy 1 Diseases 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 208000025174 PANDAS Diseases 0.000 description 1
- 208000021155 Paediatric autoimmune neuropsychiatric disorders associated with streptococcal infection Diseases 0.000 description 1
- 240000000220 Panda oleosa Species 0.000 description 1
- 235000016496 Panda oleosa Nutrition 0.000 description 1
- 241000728173 Sarima Species 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000008485 antagonism Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000012384 transportation and delivery Methods 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Evolutionary Computation (AREA)
- Strategic Management (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Operations Research (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Game Theory and Decision Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Development Economics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Educational Administration (AREA)
- Traffic Control Systems (AREA)
Abstract
The application relates to the field of behavior prediction, and particularly discloses a user behavior prediction method, a system and electronic equipment based on deep learning. The method can realize the prediction and adjustment of the passenger flow, help public transportation managers to optimize the vehicle dispatching, adjust the lines and shifts in advance so as to meet the passenger demands and reduce the congestion, thereby improving the efficiency, convenience and user satisfaction of the public transportation system.
Description
Technical Field
The present application relates to the field of behavior prediction, and more particularly, to a user behavior prediction method, system and electronic device based on deep learning.
Background
With the continual acceleration of global urbanization and the growing population, urban public transportation faces greater challenges. Densely populated cities need to provide efficient, sustainable and convenient public transportation services to meet the ever-increasing travel demands.
Sometimes, however, due to the number of passengers, the vehicle may not arrive on time or may not provide enough seats, and the passengers may need to stand or be crowded in the cabin. This can lead to crowding of the car or platform, narrowing of the space between passengers, and possibly uncomfortable experience for the passengers. A crowded environment may increase the sense of unsafe for passengers and may lead to crowd gathering and congestion, making getting on and off difficult. In addition, when passengers are more, the time to get on and off the vehicle increases, resulting in a delay in riding and an increase in queuing time. This may lengthen the waiting time for the passengers, especially during peak hours and busy stops, which may require waiting for multiple shifts to get on.
Thus, an optimized deep learning-based user behavior prediction scheme is desired.
Disclosure of Invention
The present application has been made in order to solve the above technical problems. The embodiment of the application provides a user behavior prediction method, a system and electronic equipment based on deep learning, which adopt an artificial intelligence technology based on a deep neural network model to acquire historical passenger arrival and arrival data, historical weather data, holiday information and traffic network topology diagrams, encode and fuse the historical data through context, extract image features of the traffic network topology diagrams, and generate passenger flow prediction results of all stations after the combination. The method can realize the prediction and adjustment of the passenger flow, help public transportation managers to optimize the vehicle dispatching, adjust the lines and shifts in advance so as to meet the passenger demands and reduce the congestion, thereby improving the efficiency, convenience and user satisfaction of the public transportation system.
According to an aspect of the present application, there is provided a user behavior prediction method based on deep learning, including:
acquiring historical passenger arrival and departure data, historical weather data, holiday information and a traffic network topological graph, wherein the historical passenger arrival and departure data comprises arrival and departure time, stations and lines;
constructing a historical traffic network joint feature matrix among the historical passenger arrival and departure data, the historical weather data, holiday information and the traffic network topological graph;
and generating a prediction result of the passenger flow of each station in a future period based on the historical traffic network joint feature matrix, and correspondingly adjusting based on the prediction result.
According to another aspect of the present application, there is provided a user behavior prediction system based on deep learning, including:
the historical traffic data acquisition module is used for acquiring historical passenger arrival and departure data, historical weather data, holiday information and traffic network topology diagrams, wherein the historical passenger arrival and departure data comprises arrival and departure time, stations and lines;
the joint characteristic construction module is used for constructing a historical traffic network joint characteristic matrix among the historical passenger arrival and arrival data, the historical weather data, holiday information and the traffic network topological graph;
And the flow prediction result generation module is used for generating the flow prediction result of each station passenger in a future period of time based on the historical traffic network joint feature matrix and correspondingly adjusting based on the prediction result.
According to another aspect of the application, there is also provided an electronic device comprising a processor and a memory, wherein the memory has stored thereon a computer executable program which, when executed by the processor, implements the deep learning based user behavior prediction method of any of claims 1 to 8.
Compared with the prior art, the user behavior prediction method, the system and the electronic equipment based on the deep learning adopt an artificial intelligence technology based on a deep neural network model, historical passenger arrival and arrival data, historical weather data, holiday information and traffic network topology diagrams are obtained, the historical data are encoded and fused through context, the traffic network topology diagrams are subjected to image feature extraction, and the traffic network topology diagrams are combined to generate passenger flow prediction results of all stations. The method can realize the prediction and adjustment of the passenger flow, help public transportation managers to optimize the vehicle dispatching, adjust the lines and shifts in advance so as to meet the passenger demands and reduce the congestion, thereby improving the efficiency, convenience and user satisfaction of the public transportation system.
Drawings
The foregoing and other objects, features and advantages of the present application will become more apparent from the following more particular description of embodiments of the present application, as illustrated in the accompanying drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and not constitute a limitation to the application. In the drawings, like reference numerals generally refer to like parts or steps.
Fig. 1 is a flowchart of a user behavior prediction method based on deep learning according to an embodiment of the present application.
Fig. 2 is a schematic architecture diagram of a deep learning-based user behavior prediction method according to an embodiment of the present application.
Fig. 3 is a flowchart of constructing a historical traffic network joint feature matrix among the historical passenger arrival and departure data, the historical weather data, holiday information and the various traffic network topology diagrams in the deep learning-based user behavior prediction method according to an embodiment of the present application.
Fig. 4 is a flowchart of feature extraction of the historical passenger arrival and departure data, the historical weather data and holiday information to obtain a historical data feature matrix in the deep learning-based user behavior prediction method according to an embodiment of the present application.
Fig. 5 is a flowchart of image feature collection on the traffic network topology map to obtain a traffic network feature matrix in the user behavior prediction method based on deep learning according to an embodiment of the present application.
FIG. 6 is a block diagram of a deep learning based user behavior prediction system according to an embodiment of the present application.
Fig. 7 is a block diagram of an electronic device according to an embodiment of the present application.
Detailed Description
Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure have been shown in the accompanying drawings, it is to be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are for illustration purposes only and are not intended to limit the scope of the present disclosure.
Fig. 1 is a flowchart of a user behavior prediction method based on deep learning according to an embodiment of the present application. Fig. 2 is a schematic architecture diagram of a deep learning-based user behavior prediction method according to an embodiment of the present application. As shown in fig. 1 and fig. 2, a user behavior prediction method based on deep learning according to an embodiment of the present application includes: s110, historical passenger arrival and departure data, historical weather data, holiday information and a traffic network topological graph are obtained, wherein the historical passenger arrival and departure data comprise arrival and departure time, stations and lines; s120, constructing a historical traffic network joint feature matrix among the historical passenger arrival and departure data, the historical weather data, holiday information and the traffic network topological graph; and S130, generating a prediction result of the passenger flow of each station in a future period of time based on the historical traffic network joint feature matrix, and correspondingly adjusting based on the prediction result.
Deep learning aims at realizing learning and understanding of data by simulating the working principle of a human brain neural network. Based on the concept of an artificial neural network (Artificial Neural Networks, ANNs), information processing and feature extraction are performed through a multi-level neuron structure. Specifically, the core of deep learning is a deep neural network, which consists of multiple hidden layers, each containing multiple neurons, that process input data and generate output by a combination of weights and activation functions. It should be appreciated that the key to deep learning is to automatically learn and optimize weights and parameters in the neural network through a large amount of training data and back propagation algorithms to achieve high level abstract and characterization learning of the data. The development of deep learning benefits from the availability of large-scale data sets, powerful computing resources, and improvements in algorithms. It has made breakthrough progress in many fields and has provided important support and impetus for the development of artificial intelligence.
User behavior prediction refers to predicting a user's possible future behavior by analyzing and modeling the user's past behavior data. Such behavior may include purchase decisions, clicks, browses, searches, reviews, shares, etc. of the user. The purpose of user behavior prediction is to infer future behavior trends and possible preferences of the user according to historical behavior patterns and features of the user so as to personalize applications such as recommendation, advertisement delivery, user subdivision, and accurate marketing.
In the public transportation field, passenger flow at different time periods and places can be predicted by analyzing the boarding and disembarking data and riding modes of historical passengers. This helps the mass transit manager optimize vehicle scheduling, adjust routes and shifts ahead of time to meet passenger demand and reduce congestion; by analyzing historical passenger traffic and congestion conditions, areas and periods of time of congestion hot spots that are likely in the future can be predicted. This helps the public transportation manager take corresponding measures such as increasing the number of vehicles, adjusting shifts, or guiding passengers to select other routes to alleviate the crowded situation.
In step S110, historical passenger arrival and departure data, historical weather data, holiday information, and a traffic network topology map are obtained, wherein the historical passenger arrival and departure data includes arrival and departure times, stations, and routes. The passenger arrival and departure data comprises the arrival and departure records of passengers at different stations, and the passenger flow change trend of different stations can be reflected. And weather has a great influence on the traveling behavior of passengers. For example, severe weather conditions may cause passengers to reduce or change travel patterns, while good weather may stimulate more people to choose to use public transportation. Furthermore, holidays are often periods of increased demand for people to travel, so the impact of holidays needs to be considered when predicting passenger traffic. By analyzing historical passenger arrival and arrival data, weather data and holiday information, periodic changes in one day, one week and one year can be found, and the influence of specific events or holidays on passenger flow can be known, the relation between weather and passenger flow can be known, the influence mode of holidays on passenger flow can be found, and the influence mode is considered in a prediction model, so that prediction accuracy is improved. The traffic network topology map describes the connection relationships and path information between different sites. By acquiring the traffic network topology diagram, the information such as the distance between stations, traffic flow distribution and the like can be known. This information is important for predicting passenger flow and optimizing vehicle scheduling, and can help determine optimal route planning and shift scheduling.
In particular, passenger ingress and egress data is typically collected by public transportation operations authorities or related data providers. The data can be acquired by means of a passenger card swiping record, a station access control system, vehicle-mounted equipment and the like. Public transportation authorities typically save passenger arrival and departure data for operational analysis and decision making. Weather data may be obtained from weather bureaus, weather sensors, weather forecasts, and the like. Meteorological data of various areas, including temperature, precipitation, wind power and the like, are usually collected and recorded by a meteorological office. The meteorological sensor can monitor meteorological conditions in real time and transmit data to the data acquisition system. In addition, weather forecast also provides weather forecast data over a period of time in the future. Holiday information may be obtained from government authorities' websites, calendars, holiday authorities, and the like. Government official websites typically release holiday schedules and related information, and calendars also annotate holiday dates. Holiday management is responsible for maintaining and distributing holiday information. The traffic network topology map may be obtained from channels such as traffic planning departments, traffic authorities, geographic information systems, and the like. These institutions typically maintain databases of geographic information about traffic networks, including information about roads, intersections, bus routes, stops, and the like. The geographic information system may provide detailed geographic data and traffic network topology.
In step S120, a historical traffic network joint feature matrix between the historical passenger arrival and departure data, the historical weather data, holiday information, and the traffic network topology is constructed. The aim of constructing the historical traffic network joint feature matrix is to fuse historical passenger arrival and departure data, historical weather data, holiday information and traffic network topology graphs together so as to comprehensively consider the influence of the historical passenger arrival and departure data, the historical weather data, the holiday information and the traffic network topology graphs on passenger flow in a prediction model. Specifically, by constructing the historical traffic network joint feature matrix, different types of data can be integrated and represented, so that richer feature information is extracted. The correlation and interaction among different factors can be better captured, and the accuracy and reliability of passenger flow prediction are further improved.
Specifically, the historical traffic network joint feature matrix can comprise features such as time, stations, number of in-out of passengers, and the like, and the features can reflect information such as passenger flow change trend, peak time period, and the like of different stations; characteristics of temperature, precipitation, wind, etc. in historical weather data that may reflect the impact of weather on passenger flow, e.g., the amount of passengers may decrease on hot days and increase on rainy days; holiday information may be converted to a binary feature that indicates whether a day is a holiday, so that the effect of the holiday on passenger traffic may be taken into account, as passenger traffic will typically vary during the holiday; nodes, edges, distances, traffic flow and other features in the traffic network topology graph can reflect information such as connection relations and traffic flow distribution among different stations.
Fig. 3 is a flowchart of constructing a historical traffic network joint feature matrix among the historical passenger arrival and departure data, the historical weather data, holiday information and the various traffic network topology diagrams in the deep learning-based user behavior prediction method according to an embodiment of the present application. Specifically, in the embodiment of the present application, as shown in fig. 3, constructing a historical traffic network joint feature matrix between the historical passenger arrival and arrival data, the historical weather data, holiday information and the traffic network topology diagrams includes: s210, carrying out feature extraction on the historical passenger arrival and arrival data, the historical weather data and holiday information to obtain a historical data feature matrix; s220, image feature collection is carried out on the traffic network topological graph to obtain a traffic network feature matrix; and S230, combining the historical data feature matrix and the traffic network feature matrix to obtain the historical traffic network combined feature matrix.
Fig. 4 is a flowchart of feature extraction of the historical passenger arrival and departure data, the historical weather data and holiday information to obtain a historical data feature matrix in the deep learning-based user behavior prediction method according to an embodiment of the present application. Specifically, in the embodiment of the present application, as shown in fig. 4, feature extraction is performed on the historical passenger arrival/arrival data, the historical weather data, and holiday information to obtain a historical data feature matrix, which includes: s310, inputting the historical passenger arrival and departure data, the historical weather data and holiday information into a historical data context encoder comprising a word embedding layer to obtain a plurality of historical passenger arrival and departure information feature vectors and a plurality of historical weather holiday feature vectors; s320, two-dimensionally arranging the characteristic vectors of the plurality of historical passenger arrival and departure information into a historical passenger arrival and departure characteristic matrix; s330, two-dimensionally arranging the plurality of historical weather vacation feature vectors into a historical weather vacation information feature matrix; and S340, fusing the historical passenger arrival and departure feature matrix and the historical weather vacation information feature matrix to obtain the historical data feature matrix.
In particular, in another possible embodiment of the present application, the historical passenger arrival/arrival data, the historical weather data, and holiday information may be feature extracted to obtain a historical data feature matrix by: 1. extracting the number features of the in-out stations from the in-out station data of passengers, calculating the number of the in-out stations in each time period, such as the total number of the in-out stations in each hour, the difference between the number of the in-out stations in each hour and the number of the out-out stations, and the like, and calculating the statistical features of the number of the in-out stations in each time period, such as the mean value, the variance, the maximum value, the minimum value and the like; 2. extracting a temperature value, a precipitation amount value and a wind power grade of each time point by using a time sequence model such as ARIMA, SARIMA, prophet; 3. converting holiday information into binary features, representing whether a day is a holiday or not; 4. combining the extracted features into a feature matrix, wherein each row represents the features of a time point, each column represents a feature, and if missing values exist, filling or interpolation can be performed.
Specifically, in step S310, the historical passenger arrival/arrival data, the historical weather data, and holiday information are input into a historical data context encoder including a word embedding layer to obtain a plurality of historical passenger arrival/arrival information feature vectors and a plurality of historical weather holiday feature vectors.
More specifically, inputting the historical passenger ingress and egress data, the historical weather data, holiday information into a historical data context encoder comprising a word embedding layer to obtain a plurality of historical passenger ingress and egress information feature vectors and a plurality of historical weather holiday feature vectors, comprising: word segmentation processing is carried out on the historical passenger arrival and departure data so as to convert the historical passenger arrival and departure data into word sequences composed of a plurality of words; mapping each word in the word sequence into a word embedding vector by using the embedding layer of the historical data context encoder comprising the word embedding layer so as to obtain a sequence of word embedding vectors; one-dimensional arrangement is carried out on the sequence of the word embedding vectors to obtain global feature vectors; calculating the product between the global feature vector and the transpose vector of each vector in the sequence of word embedding vectors to obtain a plurality of self-attention correlation matrices; respectively carrying out standardization processing on each self-attention correlation matrix in the plurality of self-attention correlation matrices to obtain a plurality of standardized self-attention correlation matrices; obtaining a plurality of probability values by using a Softmax classification function through each normalized self-attention correlation matrix in the normalized self-attention correlation matrices; and weighting each vector in the sequence of word embedded vectors by taking each probability value in the plurality of probability values as a weight to obtain the plurality of historical passenger arrival and arrival information feature vectors.
Specifically, in step S340, the historical passenger arrival/arrival feature matrix and the historical weather holiday information feature matrix are fused to obtain the historical data feature matrix. The passenger arrival and departure feature matrix contains information related to passenger flow, such as the number of arrival, the number of departure, the connection relationship between stations and the like. And the characteristic matrix of the historical weather vacation information contains information related to weather and vacation, such as temperature, rainfall, vacation type and the like. By combining the two feature matrices, the passenger flow and the weather vacation information can be combined, the diversity of the features is increased, and the factors influencing the passenger flow are more comprehensively described. Second, passenger traffic is affected by a variety of factors, including traffic networks, weather conditions, and holiday conditions. The complex relationships between these influencing factors may not be fully captured using passenger access features or weather holiday information alone. By fusing the two feature matrixes, a plurality of factors can be comprehensively considered, and a more accurate and comprehensive prediction basis is provided, so that the accuracy of passenger flow prediction is improved.
Specifically, in step S220, image feature collection is performed on the traffic network topology map to obtain a traffic network feature matrix. The traffic network topology diagram contains structural information such as roads, intersections, nodes and the like. Through image feature collection, information such as positions, shapes, connection relations, distribution, change and the like of the structures can be extracted, so that the attribute features of the traffic network are obtained, and the structural features of the traffic network are obtained. The structural features can be used for analyzing the topological structure of the traffic network, such as road density and node connection conditions, and can be used for analyzing the running state, congestion conditions and the like of the traffic network, so as to provide more information and tools for optimizing tasks such as public traffic management, traffic planning, traffic prediction and the like.
Fig. 5 is a flowchart of image feature collection on the traffic network topology map to obtain a traffic network feature matrix in the user behavior prediction method based on deep learning according to an embodiment of the present application. Specifically, in the embodiment of the present application, as shown in fig. 5, image feature collection is performed on the traffic network topology map to obtain a traffic network feature matrix, which includes: s410, the traffic network topological graph passes through a saliency network detector to obtain a traffic network characteristic graph; and S420, carrying out global pooling on the traffic network feature map along the channel dimension to obtain the traffic network feature matrix.
A saliency network detector (Salient Object Detection) is a computer vision technique for detecting salient objects or regions in an image. Salient objects generally refer to portions of an image that are noticeable, significantly different from the surrounding environment, such as bright objects, vivid colors, sharp edges, etc. In particular, saliency network detectors typically perform computation of saliency areas based on low-level features (e.g., color, texture, brightness, etc.) and high-level features (e.g., edges, contrast, etc.) of an image. It may be implemented by a machine learning method (e.g., feature-based classifier) or based on image processing algorithms (e.g., image filtering, segmentation, etc.).
In the technical scheme of the application, the saliency network can help identify important areas in the traffic topology map, such as intersections, traffic junctions and the like. These areas play a key role in traffic management and by highlighting them the structure and characteristics of the traffic network can be better understood and analyzed. In addition, in traffic flow hot spots in the traffic topology, i.e. traffic dense areas. The saliency network can also help identify to highlight the hot spots, so that the distribution and the change of traffic flow are better known, and traffic management measures such as optimizing signal lamp timing, adjusting road traffic direction and the like are adopted in a targeted manner, so that the traffic efficiency is improved, and the congestion is reduced. Second, by converting the traffic topology map into visually prominent images, traffic data is made easier to understand and analyze. This helps traffic managers and decision makers to more intuitively observe the state and changes of the traffic network, making corresponding decisions and adjustments.
Specifically, in step S420, the traffic network feature map is globally pooled along a channel dimension to obtain the traffic network feature matrix. Considering that traffic network feature graphs typically have a high dimension, especially when convolutional neural networks are used for feature extraction. In the technical scheme of the application, the feature map of each channel can be compressed into a single feature value through global pooling operation, so that the dimension of data is reduced. While different regions in the traffic topology may have different importance and influence. By global pooling, the entire network can be comprehensively considered, and important areas and features can be focused. The method is helpful for the model to better understand the structure and characteristics of the whole traffic network, and captures key information of the traffic network, so that the understanding and prediction capability of traffic flow hot spots and topological structures are improved.
In particular, in another possible implementation manner, the traffic network topology map may be subjected to image feature acquisition through a following step to obtain a traffic network feature matrix. The following is a pair of steps: firstly, preprocessing is carried out on the traffic network topological graph, such as image size adjustment, graying and other operations, so as to facilitate subsequent processing. And then, filtering the preprocessed traffic network topological graph, removing noise and enhancing image characteristics. And then, the filtered traffic network topological graph is segmented, and different areas or objects in the traffic network topological graph are extracted, so that the subsequent feature extraction is facilitated. Next, useful features are extracted from the segmented image for representing the structure and properties of the traffic network, such as texture features, shape features, edge features, etc. The extracted features are then encoded and converted into a digitized representation. Finally, the encoded features are integrated into a feature matrix, each row representing a sample, and each column representing a feature. The steps described above may be implemented using an image processing library (e.g., openCV) and a feature extraction library (e.g., scikit-image).
In step S130, based on the historical traffic network joint feature matrix, a prediction result of the passenger flow of each station in a future period of time is generated, and based on the prediction result, corresponding adjustment is performed.
Specifically, in the embodiment of the present application, based on the historical traffic network joint feature matrix, a prediction result of the passenger flow of each station in a future period of time is generated, and based on the prediction result, a corresponding adjustment is performed, including: and generating the prediction results of the passenger flow of each station by using the historical traffic network joint feature matrix through a passenger flow generator.
In the field of machine learning and deep learning, a Generator (Generator) refers to a model or algorithm that can generate new data samples that have similar distribution characteristics to training data. The generator is typically used to generate synthesized data such as images, text, audio, etc. Specifically, in deep learning, the generator is typically combined with a Discriminator (discriminant) to construct a generation antagonism network (GANs, generative Adversarial Networks). The GANs consist of a generator and a arbiter that train against each other. The goal of the generator is to generate what appears to be a true data sample, while the goal of the arbiter is to distinguish between the generated sample and the true sample. Through continuous countermeasure training, the generator and the discriminator can mutually promote, and the generator can generate more lifelike samples, and the discriminator can more accurately judge the authenticity of the samples.
In view of the fact that by the generator we can build passenger flow prediction models using different algorithms or models to accommodate different data characteristics and prediction needs. In detail, the prediction result generated by the generator may be used as a reference or an initial prediction, and then may be adjusted according to actual conditions. In practical applications, the prediction results may be affected by various factors, such as emergencies, special holidays, weather changes, etc. The initial prediction result generated by the generator may be used as a reference and then adjusted according to real-time data or experience of a domain expert to more accurately reflect the current situation. In addition, the generator may also provide interpretation and interpretability of the prediction results. By generating the prediction results, we can know how the model predicts according to the historical data and features, so as to increase the trust and the understandability of the prediction results. This is very important to traffic managers and decision makers because they need to understand the basis of the predicted results and make adjustments and decisions accordingly as needed.
In particular, in another possible implementation manner, the prediction results of the passenger flows at each station in a future period of time are generated based on the joint feature matrix of the historical traffic network, and the prediction results are adjusted accordingly according to the actual situation, which can be achieved by the following steps, for example: the feature extraction and processing are carried out on the combined feature matrix, and the pandas and numpy modules in the Python can be used for carrying out data processing, such as filling, feature scaling, time feature extraction and the like on the missing values; selecting a suitable time series prediction model, such as ARIMA or LSTM, using the historical passenger flow as input data, training the model to learn the time dependence and trend of the data; using the trained time series model to predict and generate the passenger flow in a future period of time, the predicted time interval can be selected according to the need, such as the prediction of each hour, each day or each week; the generated prediction result is adjusted according to the actual situation, real-time data feedback such as actual passenger flow, weather change and the like can be considered, and the prediction result is corrected and corrected so as to improve accuracy; the prediction results are visualized for better understanding and application by traffic managers and decision makers. The prediction results may be presented using charts, maps, or other visualization tools, and traffic management decisions and adjustments may be made in conjunction with other information.
It should be noted that those skilled in the art should know that the deep neural network model needs to be trained before the deep neural network model is applied to make the inference so that the deep neural network can implement a specific function.
Specifically, in the embodiment of the present application, the training method further includes the step of: training the historical data context encoder including the word embedding layer, the saliency network detector, and the passenger traffic generator.
More specifically, in an embodiment of the present application, the training step includes: acquiring training data, the training data comprising: training historical passenger arrival and arrival data, training historical weather data, training holiday information and training traffic network topology diagrams, and real results of passenger flows of all stations; inputting the passenger arrival and departure data of the training histories, the training historic weather data and the training holiday information into the history data context encoder containing the word embedding layer to obtain a plurality of training historic passenger arrival and departure information feature vectors and a plurality of training historic weather holiday feature vectors; two-dimensionally arranging the feature vectors of the plurality of training history passenger arrival and departure information into a training history passenger arrival and departure feature matrix; two-dimensionally arranging the training history weather vacation feature vectors into a training history weather vacation information feature matrix; fusing the training history passenger arrival and arrival feature matrix and the training history weather vacation information feature matrix to obtain a training history data feature matrix; passing the training traffic network topology map through the saliency network detector to obtain a training traffic network feature map; global pooling is carried out on the training traffic network feature map along the channel dimension to obtain a training traffic network feature matrix; combining the training historical data feature matrix and the training traffic network feature matrix to obtain a training historical traffic network combined feature matrix; calculating a priori-based feature engineering transition factor between the training historical data feature matrix and the training traffic network feature matrix; passing the training historical traffic network joint feature matrix through the passenger flow generator to obtain a generated loss function value; training the historical data context encoder including the word embedding layer, the saliency network detector, and the passenger flow generator with a weighted sum of the generated loss function value and a priori based feature engineering transition factor as a loss function value.
In particular, in the technical scheme of the application, the fact that the characteristic difference exists between the training historical data characteristic matrix and the training traffic network characteristic matrix due to the fact that the data sources and the data types of the training historical data characteristic matrix and the training traffic network characteristic matrix are different is considered, and specifically, the training historical data characteristic matrix and the training traffic network characteristic matrix come from different data sources. The training history data feature matrix comprises feature vectors of passenger arrival and arrival data, history weather data and holiday information, and the training traffic network feature matrix represents the topological structure of the traffic network. These two data sources represent different information and therefore their characteristic differences are unavoidable. Meanwhile, the training history data feature matrix and the training traffic network feature matrix are different in data type. The training history data feature matrix may comprise both continuous and discrete data, whereas the training traffic network feature matrix typically comprises discrete data. Different data types may result in different representations and distributions of their features, which in turn may result in feature differences.
And because the training history data feature matrix and the training traffic network feature matrix have own feature differences, the distribution and the representation modes of the training history data feature matrix and the training traffic network feature matrix in the feature space can be different. This means that their semantic features may have different spatial locations and distributions in the high-dimensional feature space. The operation of superposition convolutional coding may further transform and combine these features, possibly resulting in spatial convergence problems between the different features.
Matching at the feature level may alleviate this problem. By matching the feature matrix on the feature level, the difference between the training history data feature matrix and the training traffic network feature matrix can be reduced, so that the training history data feature matrix and the training traffic network feature matrix are more consistent. That is, by matching on the feature level, the performance of the passenger flow prediction model can be improved. The matched feature matrix can better capture the association information between the historical data and the traffic network, so that the model can learn and predict the passenger flow more accurately.
Specifically, due to the feature difference between the training historical data feature matrix and the training traffic network feature matrix, the space convergence of the high-dimensional feature space mapped by the semantic features brought by the superposition convolution coding needs to be matched on the feature level of the feature matrix. Based on the above, in the technical scheme of the application, the prior-based feature engineering transition factors are utilized to convert the matching problem between the feature matrixes into a training strategy optimization problem, so that an optimization technology is adopted to improve the matching and fusion degree between the feature matrixes.
Specifically, firstly, according to the structures and attributes of the training historical data feature matrix and the training traffic network feature matrix, a prior-based feature engineering transition factor strategy is designed, and feature values of different categories and dimensions are weighted and adjusted according to a certain prior rule, so that information loss and error accumulation in the superposition convolutional coding process are reduced. Furthermore, the prior-based characteristic engineering transition factors are used as loss functions to train model parameters, the prior approximate knowledge of the low-rank matrix can be used for interpolation of the displacement matrix, the matching degree between the matrices is constrained under the condition of no accurate manifold geometric relationship, the similarity and the difference degree between different matrices are calculated based on the prior-based characteristic engineering transition strategies, and optimization and correction are carried out according to a certain threshold value to improve the matching property between the characteristic matrices so as to improve the fusion degree.
Specifically, in the embodiment of the application, calculating the prior-based feature engineering transition factor between the training historical data feature matrix and the training traffic network feature matrix includes: calculating a priori-based feature engineering transition factor between the training history data feature matrix and the training traffic network feature matrix according to the following formula;
wherein M is 1 Representing the training history data feature matrix, M 2 Representing the training traffic network feature matrix,representing the addition of feature matrices by position, +.>Position-wise subtraction of the representation feature matrix, +.>The square of the Frobenius norm of the feature matrix, exp (·) represents the exponential operation of the matrix, α and λ represent the hyper-parameters, F, respectively loss Representing a priori based feature engineering transition factor.
In summary, a user behavior prediction method based on deep learning according to an embodiment of the present application is illustrated, which adopts an artificial intelligence technology based on a deep neural network model, and adopts an artificial intelligence technology based on a deep neural network model to obtain historical passenger arrival and arrival data, historical weather data, holiday information and traffic network topology map, combine the historical data through context encoding, extract image features of the traffic network topology map, and combine to generate a prediction result of passenger flow of each station. The method can realize the prediction and adjustment of the passenger flow, help public transportation managers to optimize the vehicle dispatching, adjust the lines and shifts in advance so as to meet the passenger demands and reduce the congestion, thereby improving the efficiency, convenience and user satisfaction of the public transportation system.
FIG. 6 is a block diagram of a deep learning based user behavior prediction system according to an embodiment of the present application. As shown in fig. 6, the deep learning-based user behavior prediction system 100 according to an embodiment of the present application includes: a historical traffic data acquisition module 110, configured to acquire historical passenger arrival and departure data, historical weather data, holiday information, and a traffic network topology map, where the historical passenger arrival and departure data includes arrival and departure times, stations, and routes; a joint feature construction module 120, configured to construct a historical traffic network joint feature matrix between the historical passenger arrival and arrival data, the historical weather data, holiday information, and the traffic network topology map; and a flow prediction result generating module 130, configured to generate a prediction result of the flow of each passenger at each station in a future period of time based on the historical traffic network joint feature matrix, and perform corresponding adjustment based on the prediction result.
Here, it will be understood by those skilled in the art that the specific operations of the respective steps in the above-described deep learning-based user behavior prediction system have been described in detail in the above description of the deep learning-based user behavior prediction method with reference to fig. 1 to 5, and thus, repetitive descriptions thereof will be omitted.
Exemplary electronic device
Fig. 7 is a block diagram of an electronic device according to an embodiment of the present application, as shown in fig. 7, including: at least one processor 701; a memory 702 communicatively coupled to the at least one processor; the memory 702 stores instructions executable by the at least one processor 701, and the instructions are executed by the at least one processor 701 to perform the image capturing method.
Where memory 702 and processor 701 are connected by a bus, the bus may comprise any number of interconnected buses and bridges, the buses connecting the various circuits of the one or more processors 701 and memory 702 together. The bus may also connect various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art, and therefore, will not be described any further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver may be one element or may be a plurality of elements, such as a plurality of receivers and transmitters, providing a means for communicating with various other apparatus over a transmission medium. The information processed by the processor 701 is transmitted over a wireless medium via an antenna, which in turn receives the information and transmits the information to the processor 701.
The processor 701 is responsible for managing the bus and general processing and may provide various functions including timing, peripheral interfaces, voltage regulation, power management, and other control functions. And memory 702 may be used to store information used by the processor in performing operations.
Embodiments of the present invention also relate to a computer-readable storage medium storing a computer program. The computer program implements the above-described method embodiments when executed by a processor.
That is, it will be understood by those skilled in the art that all or part of the steps in implementing the methods of the embodiments described above may be implemented by a program stored in a storage medium, where the program includes several instructions for causing a device (which may be a single-chip microcomputer, a chip or the like) or a processor (processor) to perform all or part of the steps in the methods of the embodiments described herein. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-only memory (ROM), a random access memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
It will be understood by those of ordinary skill in the art that the foregoing embodiments are specific examples of carrying out the invention and that various changes in form and details may be made therein without departing from the spirit and scope of the invention.
Claims (10)
1. A user behavior prediction method based on deep learning, comprising:
acquiring historical passenger arrival and departure data, historical weather data, holiday information and a traffic network topological graph, wherein the historical passenger arrival and departure data comprises arrival and departure time, stations and lines;
constructing a historical traffic network joint feature matrix among the historical passenger arrival and departure data, the historical weather data, holiday information and the traffic network topological graph;
and generating a prediction result of the passenger flow of each station in a future period based on the historical traffic network joint feature matrix, and correspondingly adjusting based on the prediction result.
2. The deep learning based user behavior prediction method of claim 1, wherein constructing a historical traffic network joint feature matrix between the historical passenger arrival and arrival data, the historical weather data, holiday information, and the respective traffic network topology map comprises:
performing feature extraction on the historical passenger arrival and arrival data, the historical weather data and holiday information to obtain a historical data feature matrix;
Image feature acquisition is carried out on the traffic network topological graph to obtain a traffic network feature matrix;
and combining the historical data feature matrix and the traffic network feature matrix to obtain the historical traffic network combined feature matrix.
3. The deep learning-based user behavior prediction method according to claim 2, wherein performing feature extraction on the historical passenger arrival/arrival data, the historical weather data, and holiday information to obtain a historical data feature matrix comprises:
inputting the historical passenger arrival and departure data, the historical weather data and holiday information into a historical data context encoder comprising a word embedding layer to obtain a plurality of historical passenger arrival and departure information feature vectors and a plurality of historical weather holiday feature vectors;
two-dimensionally arranging the plurality of historical passenger arrival and departure information feature vectors into a historical passenger arrival and departure feature matrix;
two-dimensionally arranging the plurality of historical weather vacation feature vectors into a historical weather vacation information feature matrix;
and fusing the historical passenger arrival and arrival feature matrix and the historical weather vacation information feature matrix to obtain the historical data feature matrix.
4. A deep learning based user behavior prediction method according to claim 3, wherein inputting the historical passenger arrival/departure data, the historical weather data, holiday information into a historical data context encoder comprising a word embedding layer to obtain a plurality of historical passenger arrival/departure information feature vectors and a plurality of historical weather holiday feature vectors comprises:
word segmentation processing is carried out on the historical passenger arrival and departure data so as to convert the historical passenger arrival and departure data into word sequences composed of a plurality of words;
mapping each word in the word sequence into a word embedding vector by using the embedding layer of the historical data context encoder comprising the word embedding layer so as to obtain a sequence of word embedding vectors;
one-dimensional arrangement is carried out on the sequence of the word embedding vectors to obtain global feature vectors;
calculating the product between the global feature vector and the transpose vector of each vector in the sequence of word embedding vectors to obtain a plurality of self-attention correlation matrices;
respectively carrying out standardization processing on each self-attention correlation matrix in the plurality of self-attention correlation matrices to obtain a plurality of standardized self-attention correlation matrices;
Obtaining a plurality of probability values by using a Softmax classification function through each normalized self-attention correlation matrix in the normalized self-attention correlation matrices;
and weighting each vector in the sequence of the word embedding vectors by taking each probability value in the plurality of probability values as a weight so as to obtain the characteristic vectors of the plurality of historical passenger arrival and arrival information.
5. The method for predicting user behavior based on deep learning according to claim 4, wherein the step of acquiring image features of the traffic network topology map to obtain a traffic network feature matrix comprises:
passing the traffic network topology map through a saliency network detector to obtain a traffic network feature map;
and carrying out global pooling on the traffic network feature map along the channel dimension to obtain the traffic network feature matrix.
6. The deep learning-based user behavior prediction method according to claim 5, wherein generating a prediction result of the passenger flow at each station in a future period of time based on the historical traffic network joint feature matrix, and performing corresponding adjustment based on the prediction result, comprises:
and generating the prediction results of the passenger flow of each station by using the historical traffic network joint feature matrix through a passenger flow generator.
7. The deep learning based user behavior prediction method of claim 6, further comprising a training step for training the historical data context encoder including a word embedding layer, the saliency network detector, and the passenger traffic generator;
wherein the training step comprises:
acquiring training data, the training data comprising: training historical passenger arrival and arrival data, training historical weather data, training holiday information and training traffic network topology diagrams, and real results of passenger flows of all stations;
inputting the passenger arrival and departure data of the training histories, the training historic weather data and the training holiday information into the history data context encoder containing the word embedding layer to obtain a plurality of training historic passenger arrival and departure information feature vectors and a plurality of training historic weather holiday feature vectors;
two-dimensionally arranging the feature vectors of the plurality of training history passenger arrival and departure information into a training history passenger arrival and departure feature matrix;
two-dimensionally arranging the training history weather vacation feature vectors into a training history weather vacation information feature matrix;
Fusing the training history passenger arrival and arrival feature matrix and the training history weather vacation information feature matrix to obtain a training history data feature matrix;
passing the training traffic network topology map through the saliency network detector to obtain a training traffic network feature map;
global pooling is carried out on the training traffic network feature map along the channel dimension to obtain a training traffic network feature matrix;
combining the training historical data feature matrix and the training traffic network feature matrix to obtain a training historical traffic network combined feature matrix;
calculating a priori-based feature engineering transition factor between the training historical data feature matrix and the training traffic network feature matrix;
passing the training historical traffic network joint feature matrix through the passenger flow generator to obtain a generated loss function value;
training the historical data context encoder including the word embedding layer, the saliency network detector, and the passenger flow generator with a weighted sum of the generated loss function value and a priori based feature engineering transition factor as a loss function value.
8. The deep learning based user behavior prediction method of claim 7, wherein calculating a priori based feature engineering transition factor between the training history data feature matrix and the training traffic network feature matrix comprises:
calculating a priori-based feature engineering transition factor between the training history data feature matrix and the training traffic network feature matrix according to the following formula;
wherein M is 1 Representing the training history data feature matrix, M 2 Representing the training traffic network feature matrix,representing the addition of feature matrices by position, +.>Position-wise subtraction of the representation feature matrix, +.>The square of the Frobenius norm of the feature matrix, exp (·) represents the exponential operation of the matrix, α and λ represent the hyper-parameters, F, respectively loss Representing a priori based feature engineering transition factor.
9. A deep learning-based user behavior prediction system, comprising:
the historical traffic data acquisition module is used for acquiring historical passenger arrival and departure data, historical weather data, holiday information and traffic network topology diagrams, wherein the historical passenger arrival and departure data comprises arrival and departure time, stations and lines;
The joint characteristic construction module is used for constructing a historical traffic network joint characteristic matrix among the historical passenger arrival and arrival data, the historical weather data, holiday information and the traffic network topological graph;
and the flow prediction result generation module is used for generating the flow prediction result of each station passenger in a future period of time based on the historical traffic network joint feature matrix and correspondingly adjusting based on the prediction result.
10. An electronic device, comprising:
a processor and a memory are provided for the processor,
wherein the memory has stored thereon a computer executable program which, when executed by the processor, implements the deep learning based user behavior prediction method of any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311648970.8A CN117610734A (en) | 2023-12-04 | 2023-12-04 | Deep learning-based user behavior prediction method, system and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311648970.8A CN117610734A (en) | 2023-12-04 | 2023-12-04 | Deep learning-based user behavior prediction method, system and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117610734A true CN117610734A (en) | 2024-02-27 |
Family
ID=89956079
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311648970.8A Withdrawn CN117610734A (en) | 2023-12-04 | 2023-12-04 | Deep learning-based user behavior prediction method, system and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117610734A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117636045A (en) * | 2023-12-07 | 2024-03-01 | 湖州练市漆宝木业有限公司 | Wood defect detection system based on image processing |
-
2023
- 2023-12-04 CN CN202311648970.8A patent/CN117610734A/en not_active Withdrawn
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117636045A (en) * | 2023-12-07 | 2024-03-01 | 湖州练市漆宝木业有限公司 | Wood defect detection system based on image processing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110164128B (en) | City-level intelligent traffic simulation system | |
| Bharadiya | Artificial intelligence in transportation systems a critical review | |
| US20210064999A1 (en) | Multi-scale multi-granularity spatial-temporal traffic volume prediction | |
| CN111582559B (en) | Arrival time estimation method and device | |
| CN102087722A (en) | Learning apparatus and method, prediction apparatus and method, and program | |
| CN112150207A (en) | Online taxi appointment order demand prediction method based on space-time context attention network | |
| CN114202120A (en) | Urban traffic travel time prediction method aiming at multi-source heterogeneous data | |
| CN110400462B (en) | Track traffic passenger flow monitoring and early warning method and system based on fuzzy theory | |
| CN117610734A (en) | Deep learning-based user behavior prediction method, system and electronic equipment | |
| CN113591380A (en) | Traffic flow prediction method, medium and equipment based on graph Gaussian process | |
| CN112071062A (en) | Driving time estimation method based on graph convolution network and graph attention network | |
| CN116824861B (en) | Method and system for scheduling sharing bicycle based on multidimensional data of urban brain platform | |
| CN117494034A (en) | Air quality prediction method based on traffic congestion index and multi-source data fusion | |
| CN112559585A (en) | Traffic space-time sequence single-step prediction method, system and storage medium | |
| CN113408786B (en) | Traffic characteristic prediction method and system | |
| KR102359902B1 (en) | Crossroads LOS Prediction Method Based on Big Data and AI, and Storage Medium Having the Same | |
| CN111008730B (en) | Crowd concentration prediction model construction method and device based on urban space structure | |
| Radi et al. | Enhanced Implementation of Intelligent Transportation Systems (ITS) based on Machine Learning Approaches | |
| CN117134978A (en) | Vehicle identity verification method and system based on local and global behavior pattern analysis | |
| US20230153742A1 (en) | Methods for shared bicycle delivery and operation area planning in smart cities and internet of things (iot) systems thereof | |
| CN117077846A (en) | Passenger flow volume prediction method, equipment, storage medium and device | |
| Xue et al. | Traffic status evolution trend prediction based on congestion propagation effects under rainy weather | |
| CN115565376A (en) | Vehicle travel time prediction method and system fusing graph2vec and double-layer LSTM | |
| CN115762147A (en) | Traffic flow prediction method based on adaptive graph attention neural network | |
| Zhou et al. | Refined taxi demand prediction with ST-Vec |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20240227 |
|
| WW01 | Invention patent application withdrawn after publication |