System and method for compaction of floating-point numbers within a dataset

Patent metadata

Field	Value
Publication number	US-12307089-B2
Application number	US-202318479024-A
Country	US
Kind code	B2
Filing date	Sep 30, 2023
Priority date	Oct 30, 2017
Publication date	May 20, 2025
Grant date	May 20, 2025

Abstract

A system and method for compaction of floating-point numbers within a dataset, comprising a pre-encoder, a data deconstruction engine, a library manager, a codeword storage, and a data reconstruction engine. A pre-encoder may receive a plurality of data sourcepackets with may contain one or more floating-point numbers and the received data sourcepackets are scanned to identify floating-point numbers and the identified floating-point numbers. Identified floating-point numbers may be pre-encoded into binary string representations which are low-distortion embeddings of real numbers into a Hamming space. The binary string representation may be indexed to indicate it represents a floating-point number before being compacted by a data deconstruction engine and library manager. The pre-encoding of floating-point numbers located within a sourcepacket enables the system to maximize the benefit of the compaction capabilities of the data deconstruction engine.

First claim

What is claimed is:

1. A system for compaction of floating-point numbers within a dataset, comprising:
a computing device comprising a processor, a memory, and a non-volatile data storage device;
a pre-encoder comprising a plurality of programming instructions stored in the memory and operable on the processor, wherein the plurality of programming instructions, when operating on the processor, causes the processor to:
receive a dataset for encoding, the dataset comprising one or more floating-point numbers;
scan the dataset to identify the one or more floating-point numbers;
for each identified floating-point number in the dataset:
pre-encode the floating-point number into a binary string representation;
replace the floating-point number with its binary string representation in the dataset to create a pre-encoded data set; and
create an index and logically link the binary string representation with the index, wherein the index indicates the binary string represents a floating-point number in the pre-encoded dataset.

2. The system of claim 1 , further comprising a data deconstruction engine comprising a second plurality of programming instructions stored in the memory and operable on the processor, wherein the second plurality of programming instructions, when operating on the processor, causes the processor to:
receive a pre-encoded dataset;
deconstruct the pre-encoded dataset into a plurality of sourceblocks; and
compact each of the plurality of sourceblocks by assigning a codeword to a reference code associated with each of the plurality of sourceblocks; and

wherein the pre-encoder is further configured to send the pre-encoded dataset to a data deconstruction engine.

3. The system of claim 1 , wherein the binary string representations are low-distortion embeddings of real numbers into Hamming space.

4. The system of claim 1 , wherein the binary string representation is a fixed-point representation.

5. The system of claim 1 , further comprising a codeword database configured to store a plurality of codewords.

6. The system of claim 1 , further comprising a data reconstruction engine comprising a third plurality of programming instructions stored in the memory and operable on the processor, wherein the third plurality of programming instructions, when operating on the processor, causes the processor to:
receive a plurality of sourceblocks;
check whether each of the plurality of sourceblocks has been logically linked to an index, wherein the presence of an index indicates the sourceblock is a binary string representation of a floating-point number; and
divide the sourceblocks that have been logically linked to an index by a fixed power of two in order to transform the sourceblock into its floating-point number form.

7. A method for compaction of floating-point numbers within a dataset, comprising the steps of:
receiving, at a pre-encoder, a dataset for encoding, the dataset comprising one or more floating-point numbers;
scanning the dataset to identify the one or more floating-point numbers;
for each identified floating-point number in the dataset:
pre-encoding the floating-point number into a binary string representation;
replacing the floating-point number with its binary string representation in the dataset to create a pre-encoded data set; and
creating an index and logically linking the binary string representation with the index, wherein the index indicates the binary string represents a floating-point number in the pre-encoded dataset.

8. The method of claim 7 , further comprising the steps of:
sending the pre-encoded dataset to a data deconstruction engine;
receiving, at the data deconstruction engine, the pre-encoded dataset;
deconstructing the pre-encoded dataset into a plurality of sourceblocks; and
compacting each of the plurality of sourceblocks by assigning a codeword to a reference code associated with each of the plurality of sourceblocks.

9. The method of claim 6 , wherein the binary string representations are low-distortion embeddings of real numbers into Hamming space.

10. The method of claim 6 , wherein the binary string representation is a fixed-point representation.

11. The method of claim 6 , further comprising a codeword database configured to store a plurality of codewords.

12. The method of claim 6 , further comprising the steps of:
receiving a plurality of sourceblocks;
checking whether each of the plurality of sourceblocks has been logically linked to an index, wherein the presence of an index indicates the sourceblock is a binary string representation of a floating-point number; and
dividing the sourceblocks that have been logically linked to an index by a fixed power of two in order to transform the sourceblock into its floating-point number form.

Assignees

• Atombeam Technologies Inc

Inventors

• Cooper Joshua
• Riahi Aliasghar
• Haddad Mojgan
• Riahi Ryan Kourosh
• Riahi Razmin
• Yeomans Charles

Classifications

• G06F3/0659

  Command handling arrangements, e.g. command buffers, queues, command scheduling · CPC title

• H03M7/6005

  Decoder aspects · CPC title

• G06F3/067

  Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS] · CPC title

• H03M7/6011

  Encoder aspects · CPC title

• G06F3/0623

  in relation to content · CPC title