Machine Learning Training via Temporal Distance Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning model training methods do not effectively focus on important training-input data elements, particularly those relevant to specific events, leading to suboptimal performance and data efficiency.
Innovation Solution
A method that determines temporal distances between training-input data elements and event time points, weighting losses and selecting data elements based on these distances to prioritize relevant data for training, allowing the model to focus on critical time periods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional machine learning training methods are used that treat all training data equally, then the training process is simple and uniform, but the model fails to focus on important time periods and achieves suboptimal performance
Solution Approach 1:
The patent applies local quality by assigning different weighting factors to different training-input data elements based on their temporal distance to event time points. Training data closer to events receive higher weights, while distant data receive lower weights. This creates a non-uniform training focus that prioritizes critical time periods, thereby improving prediction accuracy without requiring a fundamentally complex training architecture.
Solution Approach 2:
The patent changes the parameter of data importance by introducing temporal distance as a new dimension for weighting training samples. Instead of treating all samples equally, the system calculates temporal distances between training data timestamps and event timestamps, then transforms these distances into weighting factors that dynamically adjust the contribution of each sample to the loss function, enhancing model performance on event detection.
2Productivity
If all training-input data elements are used with equal importance, then data utilization is comprehensive, but data efficiency is reduced due to lack of focus on critical data
Solution Approach 1:
The patent applies local quality by assigning different weighting factors to different training-input data elements based on their temporal distance to event time points. Training data closer to events receive higher weights, while distant data receive lower weights. This creates a non-uniform training focus that prioritizes critical time periods, thereby improving prediction accuracy without requiring a fundamentally complex training architecture.
Solution Approach 2:
The patent implements partial action by emphasizing a subset of training data that is most relevant to event detection. Rather than uniformly processing all training samples, the system selectively focuses computational resources on data elements with high temporal proximity to events, achieving better data efficiency by concentrating training effort where it matters most while still maintaining comprehensive coverage through weighted inclusion of all data.
Data Source
AI summary
A method for training a machine learning model. The method includes: determining a plurality of training sequences of training-input data elements, wherein for each training sequence each training-input data element contains sensor data for a time point from a time period assigned to the training sequence in which a prespecified event takes place at least once at one or more respective event time points; determining, for each training-input data element, the temporal distance between the time point for which the training-input data element contains sensor data and one of the one or more respective event time points; and training the machine learning model depending on the determined temporal distances.