AI Model Layer Data Compression for Memory-Limited Electronics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies lack effective methods for compressing dynamic data output by artificial intelligence models, leading to memory capacity issues in devices with limited storage, particularly in mobile devices.
Innovation Solution
An electronic apparatus and method that encodes operation data from artificial intelligence model layers, stores the encoded data, and decodes it for use in subsequent layers, utilizing an encoder and decoder to manage memory efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If operation data is stored directly in memory without compression, then data accessibility and processing speed are improved, but memory capacity is exceeded in devices with limited storage
Solution Approach 1:
The patent extracts only the essential features of operation data by encoding them into compressed representations. The encoder extracts key operational characteristics while discarding redundant information, enabling storage within limited memory capacity while preserving necessary data for model operation.
Solution Approach 2:
The patent changes the parameter representation of operation data by transforming it into an encoded format with different statistical properties. The encoder modifies data parameters through compression transformations, reducing the volume while maintaining the essential information needed for AI model execution.
2Quantity of substance
If operation data is compressed through encoding, then memory capacity utilization is improved, but data processing complexity increases
Solution Approach 1:
The patent applies preliminary encoding action to operation data before it needs to be stored or processed. By pre-compressing the data into an encoded format, the system prepares the data in advance for efficient storage, reducing the burden on memory capacity while the encoding complexity is handled upfront rather than during runtime operations.
Data Source
AI summary
An electronic apparatus is provided. The electronic apparatus includes a memory configured to store one instruction or more and a processor configured to obtain output data by inputting input data to an artificial intelligence model including a plurality of layers by executing the instruction, and the artificial intelligence model is configured to output the output data based on operation through the plurality of layers and the processor is configured to encode operation data output from one of the plurality of layers and store the encoded operation data in the memory, obtain recovery data corresponding to the operation data by decoding the encoded operation data stored in the memory, and provide the obtained recovery data to another layer from among the plurality of layers.