Machine Learning Program for Urban Congestion Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for simulating and predicting urban congestion using deep learning technologies face challenges due to the large number of parameters involved, leading to increased calculation costs and potential processing issues.
Innovation Solution
A machine learning program and device that summarizes agent simulation parameters by clustering agents based on similarity, generating segment parameters representing the number of agents in each group, and using these summarized parameters to train an Attention model for efficient prediction of congestion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning technologies are used to simulate and predict urban congestion, then prediction accuracy can be improved, but calculation costs increase due to the large number of parameters involved
Solution Approach 1:
The patent segments the large set of agent parameters by classifying agents into multiple groups based on similarity of their movement attributes. Instead of processing all individual agent parameters, the system generates group parameters representing each segment, thereby reducing the number of parameters fed into the deep learning model while maintaining predictive accuracy.
Solution Approach 2:
The patent merges multiple individual agent parameters into consolidated group parameters by aggregating data within classified segments. This combining approach reduces parameter redundancy and decreases the computational burden on the deep learning model while preserving the essential movement patterns needed for accurate congestion prediction.
2Loss of information
If individual agent parameters are used for simulation, then detailed movement information is preserved, but the number of parameters increases leading to processing issues
Solution Approach 1:
The patent divides the population of agents into multiple segments based on similarity classification of their movement attributes. This segmentation allows the system to process groups of agents with similar characteristics together, reducing the total number of parameters while ensuring that distinctive movement patterns are captured through the segment structure.
Solution Approach 2:
The patent transforms individual agent parameters into group-level parameters through classification and aggregation. This parameter transformation changes the granularity from individual agent attributes to segment-level statistics, reducing parameter count while maintaining the essential information needed for accurate simulation and prediction.
Data Source
AI summary
A non-transitory computer-readable recording medium stores a machine learning program for causing a computer to execute processing including: acquiring a first parameter that represents an environment and a second parameter that represents a movement attribute of each of a plurality of moving bodies in the environment; classifying the plurality of moving bodies into a plurality of groups on the basis of the second parameter; generating a third parameter that indicates the number of moving bodies classified into each of the plurality of groups; and inputting the first parameter and the third parameter to a machine learning model to generate estimation information regarding movement of the plurality of moving bodies in the environment.