Machine Learning Model Training via General-Purpose Pre-training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional machine learning model training methods are inefficient due to the need for large amounts of training data, resulting in low training efficiency.
Innovation Solution
A method that uses a general-purpose image training set to obtain a machine learning model, adjusts model parameters based on special-purpose images and their classification labels, and iteratively refines the model until a training stop condition is met, allowing for quick training of a model with high classification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large amount of training data is used to train the machine learning model, then the classification accuracy of the model is improved, but the training time and resource consumption increase significantly
Solution Approach 1:
The patent applies preliminary action by pre-training the machine learning model on a large-scale general-purpose image dataset (such as ImageNet) before fine-tuning on the specific task dataset. This preliminary training phase allows the model to learn general image features and patterns in advance, so that when actual training on the specific task begins, the model already possesses foundational knowledge, significantly reducing the time and data needed for achieving high classification accuracy on the specific task.
2Measurement precision
If a large amount of training data is collected and processed, then the model classification accuracy is improved, but the data preparation time and computational resources increase
Solution Approach 1:
The patent performs preliminary training on a general-purpose dataset before specific task training, allowing the model to acquire general image understanding capabilities in advance. This eliminates the need to train from scratch on the specific task, dramatically improving training efficiency while maintaining high classification accuracy.
Solution Approach 2:
The patent extracts and reuses the pre-trained model weights and features from the general-purpose training phase, separating the general feature learning process from the specific task adaptation process. This extraction allows the specific task training to focus only on task-specific adjustments rather than learning all features from scratch, thereby improving productivity.
3Measurement precision
If the model is trained from scratch on specific task data, then the model achieves high accuracy on the specific task, but the training process consumes excessive time and resources
Solution Approach 1:
The patent implements preliminary action by conducting pre-training on a large-scale general-purpose image dataset before performing fine-tuning on the specific task dataset. This two-stage approach allows the model to first learn general image features and then adapt to specific task requirements, achieving high specific task accuracy with significantly reduced training time compared to training from scratch.
Solution Approach 2:
The patent applies parameter changes by adjusting training hyperparameters during the fine-tuning phase, such as using a lower learning rate to fine-tune the pre-trained model weights. This parameter adjustment allows the model to adapt to the specific task while preserving the general features learned during pre-training, achieving high accuracy efficiently.
Data Source
AI summary
This application relates to a machine learning model training method and apparatus, and an expression image classification method and apparatus. The machine learning model training method includes: obtaining a machine learning model that includes a model parameter and that is obtained through training according to a general-purpose image training set; determining a sample of a special-purpose image and a corresponding classification label; inputting the sample of the special-purpose image to the machine learning model, to obtain an intermediate classification result; and adjusting the model parameter of the machine learning model according to a difference between the intermediate classification result and the classification label, continuing training, and ending the training in a case that a training stop condition is met. The solutions provided in this application improve the training efficiency of the machine learning model.