Grammar Graph Pruning for Compact Side Information Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Grammar-based data compression systems face limitations in achieving optimal compression ratios due to practical constraints on the amount of side information that can be processed and transmitted, which often results in a suboptimal use of strongly correlated data.
Innovation Solution
A method is developed to generate a small but strongly correlated side information data sequence from a large candidate pool by constructing an admissible grammar, representing the data as a graph, assigning weights based on expansion frequencies, pruning edges, and iteratively updating the grammar to produce a pruned grammar that serves as effective side information for encoders and decoders.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If side information from a large candidate pool is provided to improve compression ratio, then compression performance improves, but the amount of data to be processed and transmitted increases
Solution Approach 1:
The patent extracts only the most relevant side information from a large candidate pool by pruning edges in the graph representation. Specifically, it removes edges that do not contribute to shortest paths between nodes, retaining only the essential connections needed for compression while discarding redundant information. This extraction process maintains compression effectiveness while reducing the quantity of side information that must be processed and transmitted.
Solution Approach 2:
The patent creates a simplified copy of the original data structure by constructing a graph representation and then pruning it to create a condensed side information set. This copied and simplified structure preserves the essential correlations needed for compression while using significantly less data than the full candidate pool, enabling efficient processing and transmission.
2Productivity
If more side information is processed and stored, then compression ratio improves, but time and space complexity increase
Solution Approach 1:
The patent extracts only the essential components needed for compression by identifying and removing redundant edges from the graph. This extraction of core elements reduces both the space required to store side information and the time required to process it, while maintaining the compression ratio benefits derived from having relevant side information.
Solution Approach 2:
The patent segments the side information processing into distinct phases: graph construction, edge weighting based on expansion frequencies, iterative pruning of non-essential edges, and final side information generation. This segmentation allows the system to handle large candidate pools systematically, reducing overall complexity by breaking down the processing into manageable stages that can be optimized independently.
3Productivity
If side information is transmitted to improve compression performance, then bandwidth efficiency improves, but transmission overhead increases
Solution Approach 1:
The patent extracts only the most critical side information needed for decompression by pruning edges that do not lie on shortest paths in the graph. This extraction ensures that the transmitted side information contains only the essential correlations required to reconstruct the compressed data, minimizing transmission overhead while maintaining bandwidth efficiency gains from using side information.
Data Source
AI summary
A computer-implemented method for generating side information for grammar-based data compression systems, such as YK compression systems, is described. An admissible grammar (G) for an input sequence (A(S0)) having a finite set of terminal symbols is obtained. A graph representation of the admissible grammar (G) is then constructed. An edge having a lowest weight (expansion frequency), or one not possessing the shortest distance and or shortest expanded sequence length, is then pruned from the graph representation to generate a pruned graph representation. A pruned grammar (G′) is then derived by removing the occurrence corresponding to the pruned edge from the grammar G and the starting variable (S0,i) of the pruned grammar (Gi) is then expanded to generate the side information.