Saliency Function Indexing for Stationary Dataset Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large datasets are challenging to search and analyze due to their size, necessitating a method to identify and index salient features to reduce data quantity and facilitate quicker analysis.
Innovation Solution
A method involving a saliency function is computed for data points in a stationary dataset, identifying local maxima and indexing segments corresponding to these points, which can be applied to time series signals or images, allowing for the reduction of data through salient segmentation and real-time indexing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the entire stationary dataset is stored and searched, then complete data availability is achieved, but storage requirements and search time increase substantially
Solution Approach 1:
The patent extracts only the salient features and segments from the complete stationary dataset based on computed saliency values. Instead of storing and searching the entire dataset, only the most relevant portions containing salient features are extracted and stored in the index, significantly reducing storage requirements while maintaining search effectiveness
Solution Approach 2:
The patent applies different treatment to different parts of the dataset based on their saliency. High-saliency regions (containing important features) are extracted and indexed, while low-saliency regions are discarded. This local quality approach ensures that storage and search resources are focused on the most valuable portions of the data
2Reliability
If the entire stationary dataset is searched, then comprehensive analysis is achieved, but search and analysis time increase substantially
Solution Approach 1:
The patent extracts only the salient segments from the complete dataset and stores them in an index. When a search or analysis operation is performed, only these extracted salient segments need to be processed, dramatically reducing the time required while maintaining the ability to identify important features and patterns
Solution Approach 2:
The patent performs preliminary extraction and indexing of salient features before actual search or analysis operations. This preliminary action creates a condensed representation of the data that can be quickly searched and analyzed, avoiding the need to process the entire large dataset during operational queries
3Loss of information
If all data points are processed and stored, then no information is lost, but data redundancy increases and processing efficiency decreases
Solution Approach 1:
The patent extracts only the essential information contained in salient features and segments, discarding redundant data points that do not contribute significantly to the overall meaning or value of the dataset. This extraction process eliminates information redundancy while preserving the critical information needed for effective analysis
Solution Approach 2:
The patent changes the parameter of data representation from storing all raw data points to storing condensed salient segments with associated metadata. This parameter change transforms the data structure to be more efficient for processing while maintaining the essential informational content
Data Source
AI summary
A saliency function is computed to indicate the saliency of each of a plurality of data points in a dataset. For each local maximum in the saliency function, a segment of the dataset is inserted into an index.