Shared Neural Network Module for Mobile App Storage Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing number of application programs on mobile devices leads to unnecessary waste of storage capacity due to duplicate common modules, as each program is installed and stored individually despite sharing similar functions.
Innovation Solution
An apparatus and method that compares neural networks of different application programs to identify common structures, stores only the common neural network, and uses it for processing, reducing storage needs by loading the common neural network into the main storage device when required.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all application programs are individually installed and stored on the device, then each application can function independently, but the storage capacity of auxiliary storage device and main storage device is unnecessarily consumed
Solution Approach 1:
The patent merges common neural network modules across multiple application programs into a shared storage location. Instead of storing duplicate copies of identical neural networks in each application, the system consolidates these common components into a shared repository, allowing multiple applications to reference and use the same neural network module, thereby reducing overall storage consumption while maintaining functional independence.
Solution Approach 2:
The patent creates universal neural network modules that can serve multiple different application programs simultaneously. A single neural network module is designed and stored to perform the same function across different applications, making it multi-functional. This universal module can be dynamically loaded by any application that requires its functionality, eliminating the need for each application to carry its own copy.
2Adaptability or versatility
If complete application programs are stored to ensure full functionality, then applications can run independently, but the data size stored on device increases
Solution Approach 1:
The patent segments application programs into distinct functional components, separating common neural network modules from application-specific code and data. By dividing the application structure into shareable modules and unique elements, the system can store only the necessary unique portions locally while referencing common modules from shared storage, thereby reducing the data size of individually stored applications while preserving complete functionality.
Solution Approach 2:
The patent extracts common neural network modules from individual application programs and places them in a shared storage location. This extraction process removes redundant data from each application's storage footprint while maintaining the ability to access the full functionality when needed, as the extracted modules remain accessible to all applications that require them.
3Ease of operation
If duplicate common modules are stored in each application program, then each application can operate independently, but storage capacity is wasted
Solution Approach 1:
The patent implements a reference-based copying mechanism where common neural network modules are stored once in shared storage and referenced by multiple applications. Instead of creating physical copies in each application, the system uses pointers or references that point to the original module in shared storage, allowing applications to operate independently while avoiding duplicate storage of identical data.
Data Source
AI summary
Provided is an apparatus configured to determine a common neural network based on a comparison between a first neural network included in a first application program and a second neural network included in a second application program, utilize the common neural network when the first application program or the second application program is executed.