Techniques for stochastic sequence compaction for artificial intelligence engines
AssigneeQUALCOMM Incorporated
InventorMahdi NAZM BOJNORDI
Life of this patent
Estimating the term…
Abstract
Aspects described herein relate to compacting segments of values for use by an artificial intelligence (AI) engine. Multiple approximate segments of values can be generated based at least on a code and a key. Each segment of values in a sequence of values can be compared to the multiple approximate segments of values to determine a candidate approximate segment to replace the segment of values. For each segment of values in the sequence of values, a signature corresponding to the candidate approximate segment determined for the segment of values can be stored in memory in place of the sequence of values, where the signature can indicate at least the code and the key of the candidate approximate segment.
Claims21
An apparatus for compacting segments of values for use by an artificial intelligence (AI) engine, comprising:
- one or more processors;
- one or more memories coupled with the one or more processors;
- and
- instructions stored in the one or more memories and operable, when executed by the one or more processors, to cause the apparatus to:
- generate, based at least on a code and a key, multiple approximate segments of values;
- compare each segment of values in a sequence of values to the multiple approximate segments of values to determine a candidate approximate segment to replace the segment of values;
- store, in the one or more memories and for each segment of values in the sequence of values, a signature corresponding to the candidate approximate segment determined for the segment of values, wherein the signature indicates at least the code and the key of the candidate approximate segment;
- perform operand expansion to obtain, for each signature stored in the one or more memories for the sequence of values, a corresponding segment of values based on the candidate approximate segment associated with at least the code and the key indicated by the signature;
- and
- perform, by the AI engine, an AI inference computation on an operand including the corresponding segment of values obtained by the operand expansion.
(canceled)
A computer-implemented method for compacting segments of values for use by an artificial intelligence (AI) engine, comprising:
- generating, based at least on a code and a key, multiple approximate segments of values;
- comparing each segment of values in a sequence of values to the multiple approximate segments of values to determine a candidate approximate segment to replace the segment of values;
- storing, in a memory and for each segment of values in the sequence of values, a signature corresponding to the candidate approximate segment determined for the segment of values, wherein the signature indicates at least the code and the key of the candidate approximate segment;
- performing operand expansion to obtain, for each signature stored in the one or more memories for the sequence of values, a corresponding segment of values based on the candidate approximate segment associated with at least the code and the key indicated by the signature;
- and
- performing, by the AI engine, an AI inference computation on an operand including the corresponding segment of values obtained by the operand expansion.
(canceled)
A non-transitory computer-readable medium, comprising code executable by one or more processors for compacting segments of values for use by an artificial intelligence (AI) engine, the code comprising code for:
- generating, based at least on a code and a key, multiple approximate segments of values;
- comparing each segment of values in a sequence of values to the multiple approximate segments of values to determine a candidate approximate segment to replace the segment of values;
- storing, in a memory and for each segment of values in the sequence of values, a signature corresponding to the candidate approximate segment determined for the segment of values, wherein the signature indicates at least the code and the key of the candidate approximate segment;
- performing operand expansion to obtain, for each signature stored in the one or more memories for the sequence of values, a corresponding segment of values based on the candidate approximate segment associated with at least the code and the key indicated by the signature;
- and
- performing, by the AI engine, an AI inference computation on an operand including the corresponding segment of values obtained by the operand expansion.
(canceled)
Dependencies are inferred from the claim text.