Neural Network Building System With Instruction Graphic Tags
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users face a significant learning curve when trying to automate operations with neural networks, particularly in building and customizing neural network training programs, as they need to be familiar with programming languages, limiting their ability to create new functions without deep understanding of the underlying programming language.
Innovation Solution
A neural network image identification system and building system that uses a user-friendly interface with instruction graphic tags, allowing users to select and combine program sets corresponding to different neural network layers, with built-in preset rules to ensure correct layer ordering and dimension matching, generating a neural network program without requiring direct programming language use.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If users directly write neural network training programs using programming languages, then the functionality and customization of the neural network can be achieved, but the learning time and time cost for beginners are considerably extended
Solution Approach 1:
The patent introduces a graphical user interface (GUI) as an intermediary between the user and the neural network training system. Users interact with drag-and-drop components, configuration panels, and visual workflow editors instead of directly writing code. This mediator translates user-friendly graphical operations into compiled training programs, enabling beginners to customize neural networks without learning programming languages while maintaining full functionality.
Solution Approach 2:
The patent replaces the mechanical system of manual code writing and compilation with an automated graphical interface system. The GUI automatically generates, validates, and compiles training programs based on user selections and configurations, eliminating the need for users to manually write and debug code while preserving the ability to create customized neural network training workflows.
2Ease of operation
If users are restricted to preset parameters only, then the time cost for beginners is reduced, but the ability to build new functions in the product cannot be achieved
Solution Approach 1:
The patent segments the neural network training process into modular, independently configurable components that users can select and combine through the GUI. Each component (e.g., data loading, model architecture, training parameters, evaluation metrics) is presented as a separate draggable element or configuration module, allowing users to easily assemble custom training workflows by combining predefined building blocks without writing code.
Solution Approach 2:
The patent implements a dynamic configuration system where the GUI adapts its interface and available options based on user selections and system state. As users configure different components, the interface dynamically updates to show relevant parameters and dependencies, enabling flexible function building while maintaining ease of use through context-aware guidance and automatic validation.
3Adaptability or versatility
If users manually write and compile neural network programs, then full control over the program is achieved, but errors and compilation failures increase due to incorrect layer ordering and dimension matching
Solution Approach 1:
The patent implements real-time feedback mechanisms in the GUI that automatically validate user configurations before compilation. The system provides immediate feedback on dimension compatibility between layers, correct ordering requirements, and parameter consistency, preventing erroneous programs from being generated. Visual indicators, warnings, and error messages guide users to correct configurations, ensuring program correctness while maintaining full customization capability.
Solution Approach 2:
The patent performs preliminary validation and checking of neural network configurations before the compilation stage. The GUI automatically verifies layer dimension compatibility, checks for correct architectural patterns, and validates parameter settings in advance, preventing compilation errors before they occur. This preliminary action ensures that only syntactically and semantically correct programs are generated for compilation.
Data Source
AI summary
The present disclosure relates to a neural network image identification system and a neural network building system and method used therein. The neural network building method comprises: forming a combination sequence of instruction graphic tags according to a plurality of instruction graphic tags selected by a user and displayed on a screen; combining a plurality of program sets corresponding to the plurality of instruction graphic tags in an order identical to that of contents in the combination sequence of these instruction graphic tags, to generate a neural network program; and checking whether the combination sequence of instruction graphic tags conforms to one or more preset rules before the neural network program is compiled. Therefore, the neural network image identification system is configured to identify an image to be identified captured by an image capturing device, while the neural network image identification program for identifying images by the neural network image identification system can be built by the neural network building system in accordance with needs of users.