Event-Based Signal Saliency Detection Using CNN Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current signal processing systems are inefficient as they analyze all signal content, including irrelevant information, which hampers response times and resource utilization, necessitating more efficient methods to conserve energy and enhance processing speed.
Innovation Solution
An event-based signal detection and classification system that converts input signals into the frequency domain, monitors magnitude changes exceeding a threshold, and outputs only event data for further processing, utilizing a CNN for saliency classification to filter out non-relevant data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If all signal content is analyzed, then complete signal information is processed, but processing time and resource consumption increase
Solution Approach 1:
The patent extracts only the relevant portions of the signal that contain useful information. By identifying and extracting salient events from the broadband signal while discarding irrelevant portions, the system maintains signal information completeness for important features while dramatically reducing processing time by ignoring non-informative signal segments.
Solution Approach 2:
The patent segments the continuous broadband signal into discrete events based on magnitude threshold detection. By dividing the signal into meaningful segments (events exceeding threshold) and non-events (below threshold), the system processes only the segmented relevant portions, reducing overall processing time while preserving all critical signal information.
2Reliability
If all signal content is analyzed, then comprehensive detection is achieved, but computing resource utilization decreases
Solution Approach 1:
The patent extracts and processes only the salient events from the signal that contribute to detection reliability. By removing irrelevant signal portions from processing, the system maintains comprehensive detection of important features while significantly improving computing resource utilization efficiency by eliminating wasted computational effort on non-informative data.
Solution Approach 2:
The patent applies partial action by processing only the necessary portion of the signal (events above threshold) rather than the entire signal. This selective processing achieves sufficient detection reliability for the application while dramatically improving resource utilization by avoiding excessive processing of redundant signal content.
3Productivity
If threshold-based event detection is used, then processing efficiency increases, but some signal information may be filtered out
Solution Approach 1:
The patent uses magnitude threshold as a parameter to distinguish between relevant and irrelevant signal portions. By setting an appropriate threshold level, the system achieves high processing efficiency by filtering out low-magnitude irrelevant events while preserving all high-magnitude salient events, thus maintaining signal information completeness for important features.
Solution Approach 2:
The patent applies different processing quality levels to different signal portions based on their magnitude. Events exceeding the threshold receive full processing attention (high quality), while events below the threshold are discarded (low quality). This local quality differentiation maintains information completeness for critical events while maximizing processing efficiency.
Data Source
AI summary
An event-based signal saliency detection and classification system operable to receive an input signal and convert the input signal to input data. The system transforms input data from the temporal domain into the frequency domain in a plurality of frequency domain bins. Each of the frequency domain bins are monitored for magnitude changes that meet or exceed a threshold value. Event data corresponding to an event at which time a magnitude change meets or exceeds the threshold value is detected in the frequency domain bins can be stored in an event plane. The system can output the event data a saliency-classifier convolutional neural network (CNN) to classify the event data as salient data or non-salient data and output the salient data for processing by a downstream processor to produce analysis output data.