Multi-Stage Inference Model Difficulty Calculation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inference models face challenges in achieving necessary accuracy while minimizing inference time, particularly when processing time-series data, leading to potential recognition failures due to insufficient accuracy or increased processing time.
Innovation Solution
An information processing apparatus that calculates the difficulty of inference for input data using a first-stage model and determines whether to use a later-stage model for improved accuracy, ensuring necessary inference accuracy while minimizing inference time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a machine-learned inference model with higher inference accuracy is used, then inference accuracy is improved, but inference time increases
Solution Approach 1:
The inference model is divided into multiple stages with different accuracy levels. The system segments the inference process into a first-stage model for quick processing and a second-stage model for high-accuracy processing, allowing selective application based on data difficulty
Solution Approach 2:
The system dynamically selects which inference model to apply based on the calculated difficulty of the input data. For easy data, a faster lower-accuracy model is used; for difficult data, a slower higher-accuracy model is used, making the inference process adaptive rather than static
2Loss of time
If frames are excluded from object recognition target to reduce processing time, then inference time is reduced, but recognition stability deteriorates due to missed detection targets
Solution Approach 1:
Different quality levels of inference are applied locally to different data points based on their difficulty. Easy frames are processed with a lightweight model while difficult frames are processed with a heavy-duty model, ensuring each frame receives appropriate processing resources
Solution Approach 2:
The system performs a preliminary difficulty calculation and first-stage inference before deciding whether to apply the second-stage model. This preliminary action identifies which frames require enhanced processing, preventing missed detections while avoiding unnecessary processing of easy frames
Data Source
AI summary
In order to ensure necessary inference accuracy while minimizing inference time, an information processing apparatus (1) includes: a first difficulty calculation unit (11) that calculates difficulty in inference carried out by inputting input data, constituting a time series, to a first-stage inference model among multiple-stage inference models which are configured such that use of a later-stage inference model achieves higher inference accuracy; and a first determination unit (12) that determines, on the basis of the difficulty, whether a second- or later-stage inference model will be used.