Sparse Data Structure Compression with Mode Search Approximation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional delta compression methods are ineffective for sparse data structures, such as sparse matrices in machine learning, due to large differences between neighboring numerical values, leading to unsatisfactory compression results.
Innovation Solution
A mode search approximation algorithm is employed to identify the most repeated value (mode) in the data and encode it with a significance map, using ternary tree processing for rapid mode approximation, which can operate in parallel with delta compression and preferentially use machine learning compression for successful outcomes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If conventional delta compression is used on sparse data structures, then the compression process is simple, but the compression ratio is poor due to large differences between neighboring numerical values
Solution Approach 1:
The patent changes the compression approach from delta encoding to mode-based encoding, where instead of storing differences between consecutive values, the system identifies and encodes the most frequent value (mode) and its positions. This parameter change in the encoding strategy enables effective compression of sparse data structures with large value variations.
Solution Approach 2:
The patent creates a compact representation by copying only the essential information: the mode value and a significance map indicating positions where the mode occurs. This copying strategy reduces storage requirements by avoiding redundant storage of repeated values while maintaining data integrity.
2Productivity
If mode search approximation algorithm is used to identify the most repeated value, then compression efficiency improves, but computational complexity increases
Solution Approach 1:
The patent performs preliminary mode identification and significance map creation during the compression phase. By pre-processing the data to identify the mode and its positions before actual compression, the system avoids repeated computational overhead during decompression or processing, improving overall efficiency.
Solution Approach 2:
The patent segments the compression process into distinct stages: mode identification, significance map generation, and encoding. This segmentation allows each stage to be optimized independently and enables parallel processing of different data segments, reducing overall computational complexity while maintaining compression efficiency.
Data Source
AI summary
Embodiments are generally directed to compression for compression for sparse data structures utilizing mode search approximation. An embodiment of an apparatus includes one or more processors including a graphics processor to process data; and a memory for storage of data, including compressed data. The one or more processors are to provide for compression of a data structure, including identification of a mode in the data structure, the data structure including a plurality of values and the mode being a most repeated value in a data structure, wherein identification of the mode includes application of a mode approximation operation, and encoding of an output vector to include the identified mode, a significance map to indicate locations at which the mode is present in the data structure, and remaining uncompressed data from the data structure.