18237550. MODEL TRAINING METHOD AND RELATED DEVICE simplified abstract (HUAWEI TECHNOLOGIES CO., LTD.)

From WikiPatents
Jump to navigation Jump to search

MODEL TRAINING METHOD AND RELATED DEVICE

Organization Name

HUAWEI TECHNOLOGIES CO., LTD.

Inventor(s)

Jianzhong He of Beijing (CN)

Shuaijun Chen of Shenzhen (CN)

Xu Jia of Shenzhen (CN)

Jianzhuang Liu of Shenzhen (CN)

MODEL TRAINING METHOD AND RELATED DEVICE - A simplified explanation of the abstract

This abstract first appeared for US patent application 18237550 titled 'MODEL TRAINING METHOD AND RELATED DEVICE

Simplified Explanation

The patent application describes an application that trains models in the field of artificial intelligence. It considers various factors to determine a loss used for updating model parameters, resulting in a neural network with strong generalization capabilities.

  • The method involves obtaining source domain images related to a target domain image and predicting labels for these images using a to-be-trained model.
  • A loss is calculated based on the difference between the predicted labels, indicating the disparity between them.
  • The model's parameter is updated using this loss, resulting in a neural network with improved performance.

Potential Applications

  • This technology can be applied in various fields that utilize artificial intelligence, such as computer vision, natural language processing, and speech recognition.
  • It can be used in autonomous vehicles to improve object recognition and decision-making capabilities.
  • In healthcare, it can enhance medical image analysis and diagnosis accuracy.

Problems Solved

  • The method addresses the challenge of improving the generalization capability of neural networks by considering comprehensive factors in the loss calculation.
  • It tackles the problem of overfitting, where a model performs well on training data but fails to generalize to new, unseen data.
  • By updating model parameters based on the loss, it helps optimize the model's performance and accuracy.

Benefits

  • The method leads to neural networks with strong generalization capabilities, allowing them to perform well on unseen data.
  • It improves the accuracy and performance of AI models, leading to better results in various applications.
  • By addressing overfitting, it helps prevent models from making incorrect predictions or decisions based on biased training data.


Original Abstract Submitted

This application provides a model training method in the artificial intelligence field. In a process of determining a loss used to update a model parameter, factors are comprehensively considered. Therefore, an obtained neural network has a strong generalization capability. The method in this application includes: obtaining a first source domain image associated with a target domain image and a second source domain image associated with the target domain image; obtaining a first prediction label of the first source domain image and a second prediction label of the second source domain image through a first to-be-trained model; obtaining a first loss based on the first prediction label and the second prediction label, where the first loss indicates a difference between the first prediction label and the second prediction label; and updating a parameter of the first to-be-trained model based on the first loss, to obtain a first neural network.