Angular Rotation Signal Augmentation for Direction Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data augmentation techniques destroy the transfer characteristics of original training data, making them unsuitable for models estimating the direction of arrival of acoustic signals.
Innovation Solution
An angular rotation operation is performed on the first observation signal to generate a second observation signal from a different direction, which is then added to the training data without destroying the original transfer characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If conventional data augmentation techniques are used to increase training data, then the number of training data is increased, but the transfer characteristics of the original training data are destroyed
Solution Approach 1:
The patent applies parameter changes by rotating the observation signal in the angular domain (azimuth and elevation angles) to generate augmented training data. This transformation preserves the transfer characteristics because it only changes the spatial parameters while maintaining the underlying signal structure and relationships, thus resolving the contradiction between increasing data quantity and preserving data quality
Solution Approach 2:
The patent creates synthetic training data by copying and transforming existing observation signals through angular rotation operations. The augmented data is generated by applying rotation matrices to the original signals, creating new training examples that maintain the essential characteristics of the original data while increasing the overall dataset size
2Quantity of substance
If conventional data augmentation techniques are used, then the number of training data is increased, but the direction of arrival estimation accuracy is compromised
Solution Approach 1:
The patent changes the angular parameters (azimuth and elevation) of the observation signals to generate augmented data. This parameter transformation is specifically designed for direction of arrival estimation, as it creates training examples with different spatial directions while preserving the signal's transfer characteristics, thereby improving estimation accuracy without compromising data integrity
Data Source
AI summary
An input of a first observation signal corresponding to an incoming signal from a first direction is received, an angular rotation operation of the first observation signal is performed to obtain a second observation signal corresponding to an incoming signal from a second direction that is different from the first direction and the second observation signal is added to a set of training data.