Binary Entropy Coding via Relative-Entropy Source Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing binary entropy coding methods face inefficiencies due to overhead in parallel encoding and two-codeset variable-to-variable encoding, particularly when dealing with a large number of sources, leading to increased computational demands and less efficient secondary codewords usage.
Innovation Solution
The method involves selecting a new probability set and mapping each symbol's probability from a predefined set to a new set, reducing the number of probabilities and thus the overhead, by using a mapping algorithm that minimizes relative entropy, which improves encoding and decoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If parallel encoding with multiple sources is used, then coding efficiency is improved, but overhead information about subsequence lengths increases
Solution Approach 1:
The patent merges multiple probability sources into a single unified probability distribution. Instead of maintaining separate probability models for different symbol types (leading to multiple sources and overhead), the invention combines them into one integrated model that adapts to all symbol types, thereby eliminating the overhead of transmitting multiple probability sets while maintaining coding efficiency.
Solution Approach 2:
The patent creates a universal probability model that handles multiple symbol types and contexts within a single framework. This universal model replaces the need for multiple specialized probability sources, allowing the system to adapt to different symbol types without requiring separate probability sets, thus reducing overhead while preserving coding efficiency.
2Measurement precision
If a large number of sources with many probabilities are used, then coding precision is improved, but computational complexity increases
Solution Approach 1:
The patent changes the parameter representation from multiple discrete probability values across multiple sources to a single unified probability distribution with adaptive parameters. This parameter transformation maintains the precision needed to distinguish different symbol probabilities while reducing the computational burden of managing and updating multiple separate probability models.
Solution Approach 2:
By merging multiple probability sources into one unified model, the patent reduces computational complexity while preserving coding precision. The unified model consolidates the probability information that would otherwise be distributed across multiple sources, eliminating redundant computations and simplifying the overall processing while maintaining the ability to precisely encode different symbol types.
3Adaptability or versatility
If secondary codeset is used for flush events, then incomplete subsequences are handled, but encoding efficiency decreases
Solution Approach 1:
The patent inverts the traditional approach by not requiring special secondary codes for flush events. Instead of using a separate less-efficient codeset to handle incomplete subsequences, the unified probability model naturally handles all symbol types including flush events through its adaptive nature, eliminating the need for special-case handling and maintaining encoding efficiency while preserving adaptability.
Data Source
AI summary
Methods and systems are described for combining sources in a data compression system. In a system in which a context model results in the production of multiple sources or symbols, each source being associated with a probability estimate, sources may be combined by defining another, reduced size, set of probabilities. The new set of probabilities may be a subset of the predefined set of probabilities provided by the context model. Minimizing relative entropy may be a basis upon which to define a mapping of predefined probabilities to the new set of probabilities. An input sequence that was modeled using the context model may then be entropy encoded and entropy decoded using the new set of probabilities based on a mapping between the new probabilities and the predefined probabilities.