Regression-Classification Model Fusion for Age Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning models face challenges in achieving high processing capacity and accuracy for tasks like face-image-based age estimation, particularly due to low training effectiveness and uneven sample distribution across age groups.
Innovation Solution
A training method that combines regression and classification machine learning models, using a weighted sum of loss functions to improve processing capability, where a regression model extracts feature maps and a classification model determines membership probabilities, addressing the issue of uneven sample distribution through focal loss and Mae loss functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single regression machine learning model is used for age estimation, then the model structure is simple, but the processing capability and accuracy are insufficient
Solution Approach 1:
The patent combines a regression machine learning model and a classification machine learning model into a unified training framework. The regression model extracts feature maps for continuous age estimation, while the classification model processes the same feature maps for discrete age group classification. Both models share the same input and are trained simultaneously using a combined loss function, thereby improving age estimation accuracy without requiring separate independent systems.
2Ease of manufacture
If traditional training methods are used with uniform loss functions, then the training process is simple, but the training effectiveness is low due to uneven sample distribution
Solution Approach 1:
The patent applies different loss functions to different aspects of the training process to address uneven sample distribution. A focal loss function is used for the classification model to emphasize hard samples and underrepresented age groups, while a MAE (Mean Absolute Error) loss function is used for the regression model. This localized application of different loss strategies improves training effectiveness without overcomplicating the overall training framework.
Solution Approach 2:
The patent introduces a weighting mechanism in the combined loss function that dynamically adjusts the contribution of classification loss and regression loss during training. By changing the weight parameters, the training process can adapt to different data distributions and prioritize learning from challenging samples, thereby improving training effectiveness while maintaining a relatively simple training procedure.
Data Source
AI summary
The invention relates to a machine learning model training method and device and electronic equipment, and relates to the technical field of artificial intelligence. The training method includes the following steps: inputting an image sample into a regression machine learning model, extracting a feature map of the image sample by utilizing the regression machine learning model, and determining an identification result of the image sample according to the feature map; inputting the feature map into a classification machine learning model, and determining the membership probability of the image sample belonging to each classification by using the classification machine learning model according to the feature map; calculating a first loss function according to the recognition result and the labeling result of the image sample, and calculating a second loss function according to the membership probability and the labeling result of the image sample; and training a regression machine learning model by using the first loss function and the second loss function.