Trajectory Classification Training with Resampling and Abnormal Data Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for analyzing trajectory data from intelligent electronic devices lack efficiency in preprocessing and feature extraction, leading to inaccurate classification models due to abnormal data and inconsistent sampling intervals.
Innovation Solution
A method and apparatus for training a trajectory classification model that preprocesses data by deleting abnormal entries and resampling at consistent intervals, computing features based on temporal and spatial characteristics, and adjusting machine learning model parameters for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If trajectory data is used directly for model training without preprocessing, then the training process is simpler and faster, but the classification accuracy deteriorates due to abnormal data and inconsistent sampling intervals
Solution Approach 1:
The patent applies preliminary action by performing data preprocessing before model training. Specifically, it deletes abnormal trajectory data points and resamples data to ensure consistent time intervals before feeding data into the classification model. This preliminary processing eliminates data quality issues that would otherwise degrade classification accuracy, while the systematic approach keeps the added complexity manageable.
2Reliability
If trajectory data with inconsistent sampling intervals is used directly, then data processing is simpler, but the reliability of the classification model deteriorates
Solution Approach 1:
The patent applies parameter changes by modifying the time interval parameter of trajectory data through resampling. It identifies abnormal data points where time intervals between consecutive points are inconsistent, then resamples the data to enforce uniform time intervals. This parameter standardization improves model reliability by ensuring consistent temporal features, while the automated resampling process maintains reasonable processing efficiency.
3Manufacturing precision
If abnormal trajectory data is included in training, then more data is available for training, but the classification precision deteriorates due to noise and errors
Solution Approach 1:
The patent applies the taking out principle by extracting and removing abnormal trajectory data points from the training set. It identifies abnormal points based on criteria such as inconsistent time intervals and spatial anomalies, then deletes these points before training. This extraction of harmful data elements improves classification precision by eliminating noise, while the remaining clean data provides sufficient training samples for robust model learning.
Data Source
AI summary
Embodiments of the present disclosure relate to a method for training a trajectory classification model. The method includes: acquiring trajectory data; computing a trajectory feature of the trajectory data based on a temporal feature and a spatial feature of the trajectory data, the trajectory feature comprising at least one of a curvature or a rotation angle; and training the trajectory feature to obtain the trajectory classification model. Embodiments of the present disclosure further provide an apparatus for training a trajectory classification model, an electronic device, and a computer readable medium.