Tree-Encoded Bit Mask Compression Circuits for Sparse Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern digital systems face challenges in efficiently compressing large volumes of digital data, such as those used in artificial intelligence applications, due to high resource loading, which existing compression methods fail to adequately address.
Innovation Solution
The use of tree-based encoding for bit masks in data compression circuits, where switch circuits and logic circuits selectively couple and decode bit mask sets to reduce the number of bits required to specify locations of values of interest in an input vector, employing multiplexers and binary search-like strategies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If traditional compression methods are used on large volumes of digital data, then some compression is achieved, but resource loading remains high and compression ratios are insufficient
Solution Approach 1:
The patent segments the bit mask into multiple groups of bits, where each group corresponds to a specific range of values in the input vector. This segmentation allows the compression circuit to process and encode different portions of data independently, enabling more efficient compression by targeting specific value ranges with appropriate encoding strategies rather than treating the entire dataset uniformly.
Solution Approach 2:
The patent introduces a hierarchical dimension to the compression process by organizing bits into groups and levels. The tree-based encoding structure creates multiple levels of compression, where higher levels handle broader value ranges and lower levels provide finer granularity. This dimensional organization transforms the compression approach from a single-pass linear process to a multi-level hierarchical process, achieving higher compression ratios.
2Measurement precision
If more bits are used in the bit mask to specify locations of values, then more precise value selection is achieved, but data overhead increases
Solution Approach 1:
The patent applies local quality by assigning different encoding precision to different groups of bits based on their corresponding value ranges. Instead of using uniform high precision across all value locations, the circuit uses variable precision encoding where each group of bits is encoded with the minimum necessary precision for its specific range. This reduces overall data overhead while maintaining the precision needed for accurate value selection in each local region.
Solution Approach 2:
The patent changes the encoding parameter from fixed-width binary representation to variable-width tree-based encoding. By dynamically adjusting the number of bits required to represent each group based on the actual distribution and range of values, the system achieves precise value location specification with reduced data overhead. The encoding parameters adapt to the data characteristics rather than using a static allocation.
3Productivity
If complex compression algorithms are implemented to achieve higher compression ratios, then compression efficiency improves, but device complexity increases
Solution Approach 1:
The patent implements dynamic compression by using a tree-based encoding structure that can adapt its encoding depth and granularity based on the input data characteristics. The compression circuit dynamically selects which groups of bits to compress and at what level of detail, rather than applying a fixed complex algorithm to all data. This dynamic approach achieves high compression efficiency while keeping the circuit architecture relatively simple and flexible.
Solution Approach 2:
The patent introduces an intermediary tree-based encoding layer between the raw input vector and the compressed output. This intermediary structure organizes the data in a hierarchical manner that simplifies the compression process, breaking down complex compression tasks into simpler sub-tasks handled at different tree levels. The intermediary encoding structure acts as a mediator that transforms the data into a form that is more amenable to efficient compression with simpler circuitry.
Data Source
AI summary
Embodiments of the present disclosure include techniques for compressing data using a tree encoded bit mask that may result in higher compression ratios. In one embodiment, an input vector having a plurality of values is received by a first plurality of switch circuits. Selection of the input values is controlled by sets of bits from the bit mask. The sets of bits specify locations of portions of the input vector where particular value of interest reside. The switch circuits output multiple values of the input vector, which include the particular value of interest. A second stage of switch circuits is controlled by logic circuit that detects values on the outputs of the first stage of switch circuits and outputs the values of interest. In some embodiments, the values of interest may be non-zero values of a sparse input vector, and the switch circuits may be multiplexers.