Multi-Model Inference for Higher-Quality Training Data Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image analysis systems using machine learning require large amounts of training data, which are labor-intensive to generate and can result in low-quality training data due to the accuracy limitations of existing training data generation units.
Innovation Solution
A generation device that executes a program to input analysis target data to multiple inference models, determine a specific inference result, and generate a training data set based on these results, improving the quality of the training data set.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If training data is generated by adding manual analysis results to images, then the training data can be created, but it requires a lot of man-hours to generate the training data
Solution Approach 1:
The system uses automated inference models to generate training data labels independently without manual analysis. The determination device automatically selects appropriate inference results from multiple models to create training data, eliminating the need for human analysts to add labels to images.
Solution Approach 2:
The patent replaces the mechanical process of manual analysis with an automated determination device that uses multiple inference models. This substitution transforms the labor-intensive manual labeling process into an automated computational process that rapidly generates training data.
2Productivity
If a training data generation unit with low accuracy is used, then training data can be generated quickly, but the quality of the generated training data deteriorates
Solution Approach 1:
The system segments the training data generation process into multiple independent inference models, each specializing in different aspects of analysis. By dividing the work among multiple models and then determining the best results, the system achieves both high speed and high quality training data generation.
Solution Approach 2:
The determination device dynamically selects inference results based on changing parameters such as confidence levels, model performance metrics, and data characteristics. This adaptive selection process ensures that only high-quality inference results are used to generate training data, maintaining quality while preserving speed.
3Manufacturing precision
If multiple inference models are used to generate training data, then the quality of the training data set can be improved, but the device complexity increases
Solution Approach 1:
The patent merges multiple inference models into a unified determination device that coordinates their outputs. By combining the strengths of multiple models and using a centralized determination mechanism to select the best results, the system achieves high training data quality while managing complexity through integration.
Solution Approach 2:
The determination device acts as an intermediary between multiple inference models and the training data generation process. It receives outputs from various models, evaluates their quality, and selects appropriate results for training data creation, thereby managing the complexity of multiple models without compromising output quality.
Data Source
AI summary
A generation device executes an inference process of inputting analysis target data to a plurality of inference models and outputting a plurality of inference results related to an object included in the analysis target data from the plurality of inference models, a determination process of determining, based on the plurality of inference results, a specific inference result from the plurality of inference results output by the inference process, and a generation process of generating a training data set including the specific inference result determined by the determination process and the analysis target data.