Tile-Based Data Compression for Direct Evaluation Access
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data compression methods require decompression for evaluation, which is slow and inefficient, especially for large multi-dimensional data sets that exceed computer storage capacity, leading to storage and processing challenges.
Innovation Solution
A method that breaks down data into d-dimensional hyperquaders, constructs approximating functions for each hyperquader, and determines error thresholds to compress data without decompression, allowing for rapid evaluation and storage of high-dimensional data with minimal hardware requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If conventional compression programs (zip etc.) are used to compress data, then data size is reduced, but data cannot be operated on quickly and requires decompression before use
Solution Approach 1:
The patent segments the data domain into d-dimensional hyperquaders (tiles) and constructs piecewise approximating functions on each tile. This segmentation allows the compressed data to be evaluated efficiently by only computing the relevant tile containing the query point, eliminating the need for full decompression and enabling fast operations on compressed data.
2Volume of stationary object
If data is compressed using conventional methods, then storage space is reduced, but decompression is required before evaluation which consumes additional time and resources
Solution Approach 1:
The patent performs preliminary action by constructing piecewise approximating functions on segmented tiles during the compression phase. The compressed data stores parameters of these pre-construction approximating functions, enabling direct evaluation without time-consuming decompression operations, thus eliminating decompression time while maintaining storage efficiency.
3Ease of operation
If the data is decompressed for evaluation, then complete data access is enabled, but the decompressed data may not fit into computer main memory
Solution Approach 1:
The patent extracts only the essential parameters of the approximating functions and tile definitions during compression, storing them in a compact form. During evaluation, only the specific tile containing the query point needs to be accessed and evaluated, extracting minimal data from storage without requiring full decompression, thus enabling complete data access within limited memory capacity.
4Quantity of substance
If models are simplified to reduce data volume, then storage requirements are reduced, but data accuracy is compromised
Solution Approach 1:
The patent applies local quality by constructing piecewise approximating functions on locally segmented tiles rather than using a global simplified model. Each tile can have its own approximating function tailored to local data characteristics, maintaining high accuracy in each region while keeping the overall data volume reduced through selective representation of only essential parameters.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
The method involves subdividing superior tiles into subordinate tiles in a dividing step (3). An error of an approximate function is determined compared to data in an error determination step (2). The data specifying the subordinate tiles and the function is set as compressed data when the error falls below a threshold. A part of the subordinate tiles is set as the superior tiles when the error is larger than the threshold in a decision step. The dividing and approximation steps are performed for the superior tiles. The determination and decision steps are performed for the subordinate tiles. Independent claims are also included for the following: (1) a method for storing and/or transmitting data with reduced memory demand and/or transfer capacity demand (2) a method for reading data from a memory and/or for receiving data (3) a compression device for compressing data (4) an evaluation device for evaluating compressed data.