Unlock instant, AI-driven research and patent intelligence for your innovation.

Calculation Method of Trajectory Likelihood Probability of Recurrent Neural Network Based on Output State Restriction

A cyclic neural network and computing method technology, applied in neural learning methods, biological neural network models, etc., can solve problems such as weak model expression ability, inability to get rid of Markov properties, and few model parameters.

Active Publication Date: 2021-04-30
FUDAN UNIV
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] There are also some works that use the model of reverse reinforcement learning to probabilistically model the trajectory. The main problem of these methods is that using Markov decision process for modeling and using the model of reverse reinforcement learning, although it can capture the future information, but still cannot get rid of the Markov property; the model using maximum entropy reverse reinforcement learning, although it can capture long-distance dependencies, is constrained by the lack of model parameters and the poor expressiveness of the model.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Calculation Method of Trajectory Likelihood Probability of Recurrent Neural Network Based on Output State Restriction
  • Calculation Method of Trajectory Likelihood Probability of Recurrent Neural Network Based on Output State Restriction
  • Calculation Method of Trajectory Likelihood Probability of Recurrent Neural Network Based on Output State Restriction

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0038] The present invention will be described below in conjunction with specific examples:

[0039] 1. Training neural network model parameters based on historical trajectory data

[0040] (1) Define a single-layer simple cyclic neural network N, and the corresponding parameter is W N ={W hh ,W xh ,b}, each element is initialized from a uniform distribution of [-α,α], where α is a preset constant, such as 0.03. Then the feed-forward calculation function of the neural network is N(v i , h; W N ) = σ(W xh v i +W hh h+b), where σ() is a nonlinear activation function, which may be defined as the hyperbolic tangent function tanh().

[0041] (2) For each road r in the road network i , define the corresponding I-dimensional (such as 100) word embedding as v i , initialize each element of the word vector element from a uniform distribution of [-α,α].

[0042] (3) Construct a mask matrix M, where M[i,j] is 1 if and only if r i with r j Adjacent in the road network, otherw...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention belongs to the technical field of trajectory calculation, in particular to a method for calculating the likelihood probability of a trajectory of a recurrent neural network based on output state constraints. The steps of the method of the present invention include: in the training stage, the cyclic neural network limited by the output state is used to model the trajectory, and the model parameters are trained according to the historical trajectory data; in the online query stage, each road section of the input trajectory is analyzed according to the trained model Predict the probability of moving to the next road segment. The invention utilizes the powerful expressive ability of the deep neural network to learn long-distance dependencies that cannot be captured based on the Markov model, thereby modeling trajectory data more accurately.

Description

technical field [0001] The invention belongs to the technical field of trajectory calculation, and in particular relates to a calculation method of the trajectory likelihood probability of a cycle neural network based on output state limitation. Background technique [0002] The popularity of mobile GPS devices has promoted the development of location-based services, and trajectory calculation has also emerged as the times require. With the acquisition of a large amount of trajectory data, it is possible to establish a probability model for these trajectory data. Trajectory model is a very important and fundamental problem in trajectory calculation. Probabilistic modeling of trajectories can help solve application problems related to driving paths, such as route recommendation, road condition prediction, trajectory prediction, frequent pattern mining, anomaly detection, etc., and has very large application scenarios. Existing trajectory probability modeling techniques are ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06N3/08
CPCG06N3/08G06N3/084
Inventor 孙未未吴昊
Owner FUDAN UNIV